Enabling Tools and Techniques for Organic Synthesis: A Practical Guide to Experimentation, Automation, and Computation 9781119855637

Citation preview

Enabling Tools and Techniques for Organic Synthesis

Enabling Tools and Techniques for Organic Synthesis A Practical Guide to Experimentation, Automation, and Computation

Edited by

Stephen G. Newman

Department of Chemistry & Biomolecular Sciences University of Ottawa Ottawa, Canada

Copyright © 2023 by John Wiley & Sons, Inc. All rights reserved. Published by John Wiley & Sons, Inc., Hoboken, New Jersey. Published simultaneously in Canada. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning, or otherwise, except as permitted under Section 107 or 108 of the 1976 United States Copyright Act, without either the prior written permission of the Publisher, or authorization through payment of the appropriate per-­copy fee to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, (978) 750-­8400, fax (978) 750-­4470, or on the web at www.copyright. com. Requests to the Publisher for permission should be addressed to the Permissions Department, John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, (201) 748-­6011, fax (201) 748-­6008, or online at http://www.wiley.com/go/permission. Trademarks: Wiley and the Wiley logo are trademarks or registered trademarks of John Wiley & Sons, Inc. and/or its affiliates in the United States and other countries and may not be used without written permission. All other trademarks are the property of their respective owners. John Wiley & Sons, Inc. is not associated with any product or vendor mentioned in this book. Limit of Liability/Disclaimer of Warranty: While the publisher and author have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives or written sales materials. The advice and strategies contained herein may not be suitable for your situation. You should consult with a professional where appropriate. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages. For general information on our other products and services or for technical support, please contact our Customer Care Department within the United States at (800) 762-­2974, outside the United States at (317) 572-­3993 or fax (317) 572-­4002. Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic formats. For more information about Wiley products, visit our web site at www.wiley.com. Library of Congress Cataloging-­in-­Publication Data Applied for ISBN: 9781119855637, ePDF: 9781119855651, epub:9781119855644, oBook:9781119855668 Cover Design: Wiley Cover Image: © Sergey Tarasov/Adobe Stock Photos Set in 9.5/12.5pt STIXTwoText by Straive, Pondicherry, India

v

Contents List of Contributors xv Preface xix 1 1.1 1.1.1 1.1.2 1.1.3 1.2 1.3 1.3.1 1.3.1.1 1.3.1.2 1.3.1.3 1.3.1.4 1.3.1.5 1.3.1.6 1.3.2 1.3.2.1 1.3.2.2 1.3.2.3 1.3.2.4 1.3.2.5 1.3.2.6 1.3.2.7 1.3.2.8 1.3.3 1.3.4 1.4 1.4.1

Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis 1 Pablo Díaz-Kruik, David Lim, and Francesca Paradisi Glossary 1 Introduction 1 Enzymes – the Green and Sustainable Way of the Future 1 Enzymatic and Organic Catalysis Are Not too Different from Each Other 3 Enzymes 101 4 When Should I Choose an Enzyme over a Chemical Catalyst? 4 Key Considerations for Running Biocatalytic Reactions 6 Dispelling Myths 6 Enzymes Are Not Safe to Use 7 Enzymes Are Not as Readily Available as Chemical Catalysts 7 Enzymes Are Seldom Useful Due to Their Limited Substrate Scope 7 The Cost of Enzyme Production Is Very High 8 Enzymes Are Functionally Unstable Under Organic Conditions 9 Sustainability 9 Challenges of Using Enzymes: the Need for Strict Reaction Conditions 9 Enzymes from Extremophiles 10 Solvents (and Co-solvents) 10 Concentration and Ionic Strength of the Buffer 10 pH Dependence 11 Concentration of Reactants 11 Enzyme Concentration 12 Enzyme Forms 12 Toxicity 13 What Do I Need to Start Biocatalytic Experiments in My Lab? 13 Additional Considerations 14 Transformations Catalyzed by Enzymes 15 EC – The Enzyme Commission Number 15

vi

Contents

1.4.1.1 1.4.1.2 1.4.1.3 1.4.1.4 1.4.1.5 1.4.1.6 1.4.1.7 1.4.2 1.4.2.1 1.4.2.2 1.4.2.3 1.4.2.4 1.4.2.5 1.4.2.6 1.4.3 1.5 1.5.1 1.5.1.1 1.5.1.2 1.5.1.3 1.5.1.4 1.5.2 1.5.3 1.6 1.7 1.8 ­

EC 1 – Oxoreductases  15 EC 2 – Transferases  16 EC 3 – Hydrolases  16 EC 4 – Lyases  17 EC 5 – Isomerases  18 EC 6 – Ligases  18 EC 7 – Translocases  18 Some Applications of Selected Commercially Available Enzymes  19 Horseradish Peroxidase  19 Lysozyme  19 Trypsin  20 Candida Lipase B  20 Amino Acid Dehydrogenase  20 Glycosidases  21 Engineered (Unnatural) Reactions  21 ­New Trends and Technologies in Biocatalysis  21 Flow Biocatalysis and New Technologies  21 What Is Flow Biocatalysis?  21 How Does Flow Biocatalysis Work?  21 When Is a Flow Process More Beneficial for a Specific Transformation?  23 Should One Implement Every Enzymatic Reaction in Flow?  23 Enzyme Engineering  24 Photobiocatalysis  25 ­Flow Chart to Biocatalysis  25 Case Study: Setting up a Biotransformation  27 ­Concluding Remarks  31 Additional Resources  31 References  31

Introduction to Photochemistry for the Synthetic Chemist  37 Stefano Protti, Davide Ravelli, and Maurizio Fagnoni Glossary  37 2.1 ­Introduction  38 2.1.1 Light to Make Your Synthesis Greener  38 2.1.2 A Way to Overcome HOMO/LUMO Interactions  39 2.2 ­How to Plan a Photochemical Synthesis  45 2.2.1 The Choice of the Solvent  45 2.2.2 Concentration of the Absorbing Species  47 2.2.3 The Reaction Vessel  48 2.2.4 Light Sources  48 2.2.4.1 Low-­Pressure Mercury Arcs  49 2.2.4.2 Medium-­ and High-­Pressure Mercury Arcs  50 2.2.4.3 Other Light Sources  50 2.2.5 From Batch to Flow Conditions  52 2

Contents

2.2.6 2.2.7 2.3 2.3.1 2.3.2 2.3.3 2.3.4 2.3.5 2.4 ­

Preparation of the Sample  54 Safety Equipment  54 ­Selected Applications of Photochemical/Photocatalyzed Reactions  55 Reactions Involving the C═C Double Bond  55 Reactions Involving the C═O Double Bond  58 Reactions Involving a Photoinduced Homolysis  60 Reactions Involving Singlet Oxygen  62 Reactions Involving a Photocatalytic Step  62 ­Conclusions  67 Acknowledgment  67 References  67

3

How to Confidently Become an Electrosynthetic Practitioner  73 Sylvain Charvet, Taline Kerackian, Camille Z. Rubel, and Julien C. Vantourout Glossary  73 Abbreviations  76 ­Introduction  77 ­General Definition of Organic Electrosynthesis  78 ­Why is Organic Electrosynthesis Used?  78 ­How is Organic Electrosynthesis Performed?  78 ­Where to Start with Electrosynthesis?  79 Selected General Reviews  79 Selected General Guides  79 ­Electrasyn 2.0 80 Machine and Consumables  80 Opening the IKA ElectraSyn 2.0 Box  80 Cell (Vial and Cap)  81 Electrodes  82 Interface  83 Hardware  83 Menus  84 How to Set Up the Cell  84 How to Start an Experiment  85 During the Reaction  88 After the Reaction  89 ­Case Study  90 Project Overview  90 Optimization of Parameters  92 Designing an Electrochemical Experiment  92 Proof of Concept  94 Optimization  94 Substrate Scope  102 ­Conclusion  103 ­References  103

3.1 3.2 3.3 3.4 3.5 3.6 3.6.1 3.6.1.1 3.6.1.2 3.6.1.3 3.6.2 3.6.2.1 3.6.2.2 3.6.3 3.6.4 3.6.5 3.6.6 3.7 3.7.1 3.7.2 3.7.2.1 3.7.3 3.7.3.1 3.7.3.2 3.8

vii

viii

Contents

4 4.1 4.1.1 4.1.1.1 4.1.1.2 4.1.1.3 4.1.2 4.1.2.1 4.1.2.2 4.1.2.3 4.1.3 4.1.3.1 4.1.3.2 4.2 4.2.1 4.2.1.1 4.2.1.2 4.2.1.3 4.2.1.4 4.2.1.5 4.2.2 4.2.2.1 4.2.2.2 4.2.2.3 4.2.2.4 4.2.2.5 4.3 4.3.1 4.3.1.1 4.3.1.2 4.3.1.3 4.3.1.4 4.3.1.5 4.3.2 4.3.3 4.4 4.4.1 4.4.2 4.4.3 4.5 ­

Flow Chemistry  107 Yosuke Ashikari and Aiichiro Nagaki Glossary  107 ­Introduction  109 What is Flow Microchemistry  109 Reaction Time Controllability  110 Fast Mixing  111 Temperature Controllability  112 Reactions Enabled by Flow Microreactors  112 Competitive Sequential Reactions  112 Reactions Mediated by Unstable Intermediates  114 Reactions Occurring at the Surface: Two-­Phase Reactions, Electrochemical Reactions, and Photoreactions  117 Further Applicability of Flow Microsynthesis  118 Scalability  118 Safety Operation  118 ­General Information for Flow Microreactors  118 Tools and Equipment for Flow Chemistry  119 Micromixer  119 Tube Reactor  120 Pump  120 Pre-­Cooling Tubes  121 PTFE Tubes  121 How to Perform Experiments  122 Selection of Reaction Conditions  122 Preparation of Reagent Solution  125 Preparation for Reactions  126 Preparation for Reaction Evaluation  128 Cleaning Up  129 ­Case Studies  129 Competitive Sequential Reaction (General Procedure)  129 Preparation  130 Experiment  132 Screening of Reaction Conditions  133 Analysis  134 Clean Up  136 Reactions Mediated by Short-­Lived Intermediates  136 Reaction Integration  139 ­Further Expertise  142 Reaction Integration  142 Chemoselective Reactions  143 Heterogeneous Catalytic Reactions  143 ­Summary and Outlook  144 References  144

Contents

5 5.1 5.1.1 5.1.2 5.1.2.1 5.1.3 5.2 5.3 5.3.1 5.3.2 5.4 5.4.1 5.4.1.1 5.4.1.2 5.4.2 5.4.3 5.5 5.5.1 5.5.2 5.5.2.1 5.5.2.2 5.5.3 5.5.4 5.5.5 5.5.6 5.5.7 5.5.7.1 5.5.7.2 5.5.7.3 5.5.8 5.6 5.7 5.8 5.9 ­ 6

6.1

Reaction Optimization Using Design of Experiments  149 Laura Forfar and Paul Murray Glossary  149 ­Introduction  151 How Do We Experiment and DoE Terminology  151 OVAT vs. DoE  153 A Simple Chemical Example  153 A Note on Error, Accuracy, and Precision  156 ­When and How Can DoE Be Used?  157 ­What Information Can I Get from a DoE and How Is It Obtained?  158 Which Factors Are Important?  159 How Are the Models Generated?  161 ­What Types of Design Are Available?  164 Screening Designs  164 Fractional Factorial Designs  165 Definitive Screening Designs  166 Designs for Optimizing Reactions  167 Response Surface Designs  167 ­The DoE Process  169 Aim and Objective  170 Selecting Factors and Ranges  171 Factors  171 Ranges  173 Selecting Responses  175 Select a Design to Answer the Objective  176 Carry Out Design and Analyze Samples  177 Check Results  178 Model Data  179 General Steps for Developing a Model  180 Wittig Reaction  181 Complementing the Design  187 Validate Predictions  189 ­Combining DoE with Other Screening and Optimization Techniques  191 ­Software  192 ­“I Tried Experimental Design But It Did Not Work”  193 ­Conclusion  194 References  195 Introduction to High-­Throughput Experimentation (HTE) for the Synthetic Chemist  197 Stephanie Felten, Michael Shevlin, and Marion H. Emmert Glossary  197 ­What Is HTE?  199

ix

x

Contents

6.2 6.2.1 6.2.1.1 6.2.1.2 6.2.1.3 6.2.1.4 6.2.2 6.2.2.1 6.2.2.2 6.2.2.3 6.2.2.4 6.2.2.5 6.2.2.6 6.3 6.3.1 6.3.2 6.3.2.1 6.3.2.2 6.3.3 6.3.4 6.3.4.1 6.3.4.2 6.3.4.3 6.3.4.4 6.3.4.5 6.3.4.6 6.3.5 6.3.5.1 6.3.5.2 6.3.5.3 6.3.5.4 6.3.5.5 6.3.6 6.3.6.1 6.3.6.2 6.3.6.3 6.3.6.4 6.3.6.5 6.3.7 6.4 6.5

­Why HTE and What Can It Achieve?  199 Commonly Perceived Barriers to Employing HTE in Synthetic Chemistry  200 Cost  200 Availability of Dedicated HTE Facilities  200 Access to Knowledge and Training  201 Perception of HTE as Antithesis of Hypothesis-­driven Research  201 Advantages of HTE Workflows vs. Traditional Reaction Setup  203 Setup Time per Reaction  203 Miniaturization and Efficient Reagent Use  203 Multivariable vs. Sequential Optimization  203 Visualizing Reactivity Patterns  204 Serendipity in Reaction Discovery  204 Avoiding Cross-­contamination  206 ­Practical Considerations and Tools for HTE  206 Outline of a Typical HTE Workflow  207 Types of HTE Designs  209 HTE for Reaction Discovery  209 HTE for Reaction Optimization  210 HTE Design Software: Tools for Building Arrays  211 HTE Reactors and Consumables  214 Reaction Blocks  214 HTE Vials  214 Reaction Blocks with Sealing Top Plate  215 Special Reactors for Photochemistry, Electrochemistry, and High-­Pressure Reactions  215 Reaction Stirring and Temperature Control  217 Consumables  219 Considerations for Experimental Setup  220 Reaction Atmosphere  220 Reagent Preparation and Dispensing  221 Storage of Preplated Reagents  223 Pipetting  224 Solvent Evaporation  225 Analysis of HTE Screens  226 Suitable Instrumentation  226 Autosampler Configurations  226 Analytical Methods  227 Internal Standards and Assay Yields  227 Data Visualization and Analysis  228 The Role of Automation and Robotics in HTE  229 ­Section Summary and Outlook  232 Case Study 1: Development of an HTE Platform for Nickel-­Catalyzed Suzuki–Miyaura Reactions  233

Contents

6.5.1 6.5.2 6.5.3 6.5.4 6.5.5 6.6 6.6.1 6.6.2 6.6.3 6.6.4 6.6.5 6.7 6.7.1 6.7.2 6.7.2.1 6.7.2.2 6.7.2.3 6.7.2.4 6.7.2.5 ­ ­ 7 7.1 7.2 7.3 7.3.1 7.3.2 7.3.3 7.3.4 7.3.5 7.3.6 7.3.7 7.3.8 7.4 7.4.1 7.4.2 7.4.3 7.4.4

Motivation  233 Design of Test Reaction and Initial Ligand Screen  233 Second Round of Ligand/Base/Solvent Screens  235 Final Platform Design  237 Validation of Platform Design  237 Case Study 2: HTE Enabled Reaction Discovery and Optimization of ­Silyl-­Triflate-­Mediated C–H Aminoalkylation of Azoles  240 Motivation  240 Reaction Discovery Plate Design  240 Ligand Screen  243 Parallel Optimization of Three Reagents  244 Base Screen  244 ­Current Challenges and the Future of HTE  247 Summary and Conclusions  247 Remaining Challenges: The Next Frontiers  248 Biphasic Reaction Mixtures  248 Flow Chemistry and HTE  248 Reaction Profiling  249 Building Machine Learning Models to Predict Reactivity  249 Addressing Future Challenges  250 Acknowledgments  250 Further Recommended Reading  250 References  250 Concepts and Practical Aspects of Computational Chemistry  259 Martin Breugst Glossary  259 ­Introduction  261 ­Hardware­ and Software Requirements for Computational Investigations  264 ­Typical Methods in Computational Organic Chemistry  265 General Aspects  265 Molecular Mechanics and Force Fields  266 Wave-­Function Methods I – Hartree–Fock Theory  267 Wave-­Function Methods II – Post-­Hartree–Fock Theory  267 Semiempirical Methods  269 Density Functional Theory  269 Dispersion-­Corrected Density Functional Theory  271 Typical Computational Times  272 ­Basis Sets Used in Computational Organic Chemistry  273 General Aspects of Basis Sets  273 Introduction to the Mathematical Formalism in Basis Sets  274 Polarization and Diffuse Functions  275 Basis Set Families  276

xi

xii

Contents

7.4.5 7.4.6 7.5 7.5.1 7.5.2 7.5.3 7.5.4 7.5.5 7.5.6 7.6 7.7 7.7.1 7.7.2 7.7.3 7.7.4 7.7.5 7.7.6 7.8 7.8.1 7.8.2 7.8.3 7.8.4 7.8.5 7.9 7.10 7.11 ­

Effective Core Potentials (Pseudopotentials)  278 The Basis Set Superposition Error (BSSE)  279 ­Typical Computational Tasks in Organic Chemistry  279 Preliminary Remarks  279 Single-­Point Calculations  281 Geometry Optimizations  281 Frequency Calculations  282 Intrinsic Reaction Coordinate (IRC) Calculations  284 Conformational Analysis  285 ­Notation of the Model Chemistry  286 ­The Diels–Alder Reaction as a Tutorial Case Study  286 General Aspects and Requirements  286 Preparing Input Files  288 Conformational Sampling – Generation of Initial Geometries  290 Geometry Optimizations of Starting Materials and Products  291 Locating the Transition States  294 Verifying the Nature of the Transition State  298 ­More Advanced Aspects  300 General Comments  300 Influence of Solvation  300 Integration Grid  302 Standard States  302 Treating Unpaired Electrons  303 ­Important and Frequently Used Keywords  304 ­Practical Considerations  304 ­Conclusions  306 References  306

8

NMR Prediction with Computational Chemistry  313 Amy T. Merrill, Wentao Guo, and Dean J. Tantillo Glossary  313 ­Introduction  314 ­Quantum-­Chemistry-­Based Computational NMR  315 Methods  315 Time/Resources for Calculations  316 Structural Considerations in Modeling  317 Geometry Optimizations  323 Calculating Isotropic Shielding Constants  324 Common Pitfalls and How to Address Them  328 Converting to Chemical Shifts  329 Calculating Coupling Constants  330 Confidence Analysis  330 Computer-­Aided Automated Approaches  332

8.1 8.2 8.2.1 8.2.1.1 8.2.1.2 8.2.1.3 8.2.1.4 8.2.1.5 8.2.1.6 8.2.1.7 8.2.2 8.2.3

Contents

8.2.3.1 8.2.4 8.2.5 8.3 ­

CASE  332 A Case Study  336 Practicing 1H and 13C Chemical Shift Prediction  338 ­Summary and Outlook  339 Key References  339 References  340

9

Introduction to Programming for the Organic Chemist  347 Jason M. Stevens ­Introduction  347 ­Better Visualizations: Communicating Structure–Data Relationships  351 ­Text Extraction: Automating Density Functional Theory Calculations  354 ­Statistical Analysis: Deriving Insight from Historical Data  357 ­Machine Learning: A Predictive Model for Deoxyfluorination  359 ­Working with Public Datasets: Identifying Reactivity Cliffs  364 ­Running Simulations: Process Greenness  367 ­Application Development: Process Mass Intensity Predictor  371 ­Machine Learning for Reaction Optimization  374 ­Executing Robotic Tasks  378 ­Autonomous Reaction Optimization  381 ­Conclusion  384 References  385

9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8 9.9 9.10 9.11 9.12 ­ 10

10.1 10.2 10.3 10.3.1 10.3.2 10.3.3 10.3.4 10.4 10.4.1 10.4.2 10.4.3 10.4.4

Machine Learning for the Optimization of Chemical Reaction Conditions  393 A. Filipa de Almeida and Tiago Rodrigues Glossary  393 ­Introduction  394 ­Prior Art and Alternative Methods for Rational Reaction Optimization  396 ­Reaction Optimization Using LabMate.ML  400 Step One: Accessing the LabMate.ML Code and Installation  401 Step Two: Initializing the Optimization Routine in LabMate.ML  402 Step Three: Iterative Optimization Routine  404 Examples  406 ­Primer on Evaluation Guidelines  408 Code and Dataset Availability  408 Retrospective Evaluation  409 Baselines and Comparing Tools  410 Prospective Evaluation  412

xiii

xiv

Contents

10.5 ­

­ utlook  414 O References  416

11

Computer-­Assisted Synthesis Planning  423 Zhengkai Tu, Itai Levin, and Connor W. Coley Glossary  423 ­Introduction to Computer-­Aided Synthesis Planning  424 Defining the Tasks and Use Cases  424 Historical Approaches to Computer-­Aided Synthesis Planning  425 The Inflection Point of CASP Methods  425 Preliminaries on Molecular Representation and Cheminformatics  426 Outline of the Rest of the Chapter  428 ­Approaches and Algorithms for Retrosynthesis  428 Data-­driven v. Expert-­Driven Programs  428 Template-­Based Approaches  429 Template-­free Approaches with Graphs and Sequences  431 Multistep Planning Algorithms  433 ­Approaches and Algorithms for Condition Recommendation and Forward Synthesis  436 Condition Recommendation Approaches  436 Forward Synthesis Approaches  437 ­Select Examples of Software Tools for CASP  439 Open-­Source Tools  439 ASKCOS  439 AiZynthFinder  440 Retro*  442 Closed-­Source Tools  443 CASP Tools for Enzymatic Catalysis  446 Practical Considerations for CASP Programs  446 Traceability to Literature Precedent  447 How to Use CASP: Command Line Versus Graphical User Interface  447 Data Privacy  448 Customization Ability  448 ­Case Studies  448 Segler et al.’s Data-­driven Program and A/B Testing Success  449 MIT’s ASKCOS Program and Robotic Synthesis Demonstration  449 Grzybowski’s Chematica/Synthia Program’s Experimental Validations and Acquisition  450 ­Conclusion  451 Key References  453 ­References  453

11.1 11.1.1 11.1.2 11.1.3 11.1.4 11.1.5 11.2 11.2.1 11.2.2 11.2.3 11.2.4 11.3 11.3.1 11.3.2 11.4 11.4.1 11.4.1.1 11.4.1.2 11.4.1.3 11.4.2 11.4.3 11.4.4 11.4.4.1 11.4.4.2 11.4.4.3 11.4.4.4 11.5 11.5.1 11.5.2 11.5.3 11.6

Index  461

xv

List of Contributors Yosuke Ashikari Department of Chemistry Faculty of Science Hokkaido University Sapporo, Japan

Pablo Díaz-­Kruik Department of Chemistry Biochemistry and Pharmaceutical Sciences Bern, Switzerland

Martin Breugst Institut für Chemie Technische Universität Chemnitz Chemnitz, Germany

Marion H. Emmert Process Research & Development Merck & Co., Inc. Rahway, NJ, USA

Sylvain Charvet Institut de Chimie et Biochimie Moléculaires et Supramoléculaires (ICMBS, UMR 5246 du CNRS) Université Lyon Villeurbanne, France

Maurizio Fagnoni PhotoGreen Laboratory Department of Chemistry University of Pavia Pavia, Italy

Connor W. Coley Department of Chemical Engineering Massachusetts Institute of Technology Cambridge, MA, USA and Department of Electrical Engineering and Computer Science Massachusetts Institute of Technology Cambridge, MA, USA

Stephanie Felten Process Research & Development Merck & Co., Inc. Rahway, NJ, USA A. Filipa de Almeida Instituto de Investigação do Medicamento (iMed) Faculdade de Farmácia Universidade de Lisboa Lisbon, Portugal

xvi

List of Contributors

Laura Forfar Paul Murray Catalysis Consulting Ltd Yate, Bristol, UK Wentao Guo Department of Chemistry University of California Davis, CA, USA Taline Kerackian Institut de Chimie et Biochimie Moléculaires et Supramoléculaires (ICMBS, UMR 5246 du CNRS) Université Lyon Villeurbanne, France Itai Levin Synthetic Biology Center Department of Biological Engineering Massachusetts Institute of Technology Cambridge, MA, USA and Department of Chemical Engineering Massachusetts Institute of Technology Cambridge, MA, USA David Lim Department of Chemistry Biochemistry and Pharmaceutical Sciences Bern, Switzerland Amy T. Merrill Department of Chemistry University of California Davis, CA, USA Paul Murray Paul Murray Catalysis Consulting Ltd Yate, Bristol, UK

Aiichiro Nagaki Department of Chemistry Faculty of Science Hokkaido University Sapporo, Japan Francesca Paradisi Department of Chemistry Biochemistry and Pharmaceutical Sciences Bern, Switzerland Stefano Protti PhotoGreen Laboratory Department of Chemistry University of Pavia Pavia, Italy Davide Ravelli PhotoGreen Laboratory Department of Chemistry University of Pavia Pavia, Italy Tiago Rodrigues Instituto de Investigação do Medicamento (iMed) Faculdade de Farmácia Universidade de Lisboa Lisbon, Portugal Camille Z. Rubel Institut de Chimie et Biochimie Moléculaires et Supramoléculaires (ICMBS, UMR 5246 du CNRS) Université Lyon Villeurbanne, France and Department of Chemistry The Scripps Research Institute La Jolla, CA, USA

List of Contributors

Michael Shevlin Process Research & Development Merck & Co., Inc. Rahway, NJ, USA Jason M. Stevens Bristol Myers Squibb Summit, NJ, USA Dean J. Tantillo Department of Chemistry University of California Davis, CA, USA

Zhengkai Tu Computational Science and Engineering Massachusetts Institute of Technology Cambridge, MA, USA Julien C. Vantourout Institut de Chimie et Biochimie Moléculaires et Supramoléculaires (ICMBS, UMR 5246 du CNRS) Université Lyon Villeurbanne, France

xvii

xix

Preface At the undergraduate level, the organic chemistry curriculum at most universities is similar. Professors emphasize the fundamental concepts necessary to ­understand how, when, and why organic molecules interact, while lab instructors familiarize students with important hands-­on aspects of carrying out experiments. Bachelor’s students can expect to finish their studies with an idea of how molecules behave, how they are made, and how technologies such as NMR and IR can be used for their characterization. Those that enter graduate school are often surprised at the breadth of powerful technologies that make advanced organic chemistry the discipline it is today. Instead of a typical stirred round-­bottomed flask, many reactions are better done using photochemical, electrochemical, or flow reactors. Computational chemistry, once reserved for dedicated experts, can now be used by organic chemists to help predict outcomes, understand selectivity, and decipher reaction mechanisms. Automation technology can be used to generate large amounts of data with limited amounts of material, and data processing software can be used to extract subtle trends. Due to their prominence in the recent literature, trainees and established ­chemists alike would benefit from gaining expertise with these technologies to be best prepared for solving the diverse synthetic challenges that come their way. However, the barrier to learning techniques without formal instruction can be high. Even if one is fortunate enough to have access to advanced training, the expert instructor may not necessarily curate the course to the needs and the background of a synthetic chemist. The primary literature and recent textbooks have similar limitations  – while there is no shortage of resources, experts generally write to other experts, and the interested organic chemist may have critical gaps in their understanding and struggle with subdiscipline-­specific jargon. The goal of this text is to help fill this gap by providing synthetic chemists with a user-­friendly starting point to initiate their journey in developing new skills and knowledge. In each of the 11 chapters, experts communicate basic information about an impactful technology in a manner accessible to a classically trained

xx

Preface

synthetic chemist. Chapters also includes a glossary of common terminology, a general introduction to the technology of interest, case-­study examples of how it may useful to synthetic chemists, a practical discussion about steps one may take to put knowledge into practice, and references to recommended further reading. The book seeks to be a go-­to resource for organic chemists at or above the graduate level that wish to expand the breadth of tools they can use to perform, analyze, and interpret chemistry experiments. After completion, the reader will be armed with the practical knowledge needed to comprehend the literature, to assess the strengths and limitations of each technique, and to begin applying modern tools to solve synthetic challenges. This will make it useful as a general resource for graduate students looking to expand their expertise, for instructors of graduate-­ level courses on advanced techniques for organic synthesis, and for industrial scientists seeking a beginner-­friendly way to expand their knowledge. The book is organized into four subsections. Chapters  1–4 describe different enabling technologies for performing chemical experiments – biocatalysis, photochemistry, electrochemistry, and flow chemistry. While none of these topics are fundamentally new, their power as a tool for organic synthesis is becoming increasingly evident. These chapters will help the reader overcome the technical barrier hindering them from comfortably replicating experiments and designing their own. Chapters 5 and 6 focus on improved approaches to select, carry out, and analyze experiments. Specifically, Chapter 5 describes a statistical approach to experimentation that can be used to understand and optimize chemical reactions. This Design of Experiments (DoE) technique is commonly employed by practicing scientists in many fields but is seldom taught to chemists. Chapter 6 describes techniques that researchers can use to get more data using less time and fewer resources. This high-­throughput experimentation (HTE) approach shows the reader how to carry out reactions in parallel and how the collected data can be interpreted to gain insights that might otherwise be missed. Chapters  7 and  8 introduce the reader to computational chemistry tools that enable molecules and reactions to be modeled in silico, providing predictions and mechanistic insight to supplement experimentation. Chapter 7 provides a general overview of the most common computational tasks that an organic chemist may want to carry out and walks the reader through a beginner-­friendly case study wherein the reactants, transition states, and products of a Diels–Alder reaction are calculated. Chapter 8 builds upon the general knowledge given in the previous chapter and describes how computational chemistry can be used to predict the NMR spectrum of organic molecules. The goal of this chapter is to put this powerful technique into the hands of experimental chemists, which should be achievable after familiarizing themselves with the simplified approach detailed throughout. Chapters 9–11 provide the reader with an introduction to programming and machine learning. Computers already play a critical role in the daily life of a synthetic chemist, and

Preface

a little bit of familiarity with modern techniques can go a long way. Chapter  9 provides a blueprint for understanding how and why a chemist may go about familiarizing themselves with programming. Chapter  10 describes a deep dive case study for using machine learning to facilitate reaction optimization, ­providing a step-­by-­step guide that a beginner may follow to use the tool and to gain confidence in harnessing other published algorithms. Chapter 11 explains how computers can facilitate the planning of multistep synthesis by suggesting synthetic routes and reaction conditions. Helpful discussions on the current tools available, how they work, and their associated strengths and weaknesses are also described. This project was only possible due to an immense amount of work by the authors who generously agreed to share their knowledge and meet the formidable task of communicating with a general audience. I am also indebted to the many students and postdoctoral fellows at the University of Ottawa that served as reviewers to help ensure that the content serves as a welcoming and beginner-­ friendly introduction to these topics that are becoming increasingly important to the modern synthetic chemists. I hope the readers agree that this goal has been met and that this marks the beginning of their journey to being a more well-­ rounded scientist capable of tackling diverse problems that come their way. Stephen G. Newman Ottawa July 2023

xxi

1

1 Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis Pablo Díaz-­Kruik, David Lim, and Francesca Paradisi Department of Chemistry, Biochemistry and Pharmaceutical Sciences, Bern, Switzerland

Glossary API  Active pharmaceutical ingredient BRENDA  A comprehensive enzyme information system CALB  Lipase B from Candida antarctica Cofactor  A non-protein chemical compound or metallic ion that is required for an enzyme’s role as a catalyst DKR  Dynamic kinetic resolution IRED  Imine reductase NAD+/NADH  Nicotinamide adenine dinucleotide Protein expression  Biological process where the protein is synthesized inside a cell Recombinant DNA  DNA scaffold that contains the protein sequence of interest TRIS  Tris(hydroxymethyl)aminomethane

1.1 ­Introduction 1.1.1  Enzymes – the Green and Sustainable Way of the Future Recent efforts by chemists to actively reduce toxic waste production and minimize costs have led to the discovery of many green and sustainable technologies. Not surprisingly, the use of enzymes, Nature’s catalysts, has seen a major resurgence in academic and industrial interest over the past decade  – not only for their ­sustainability and natural activities but for engineering them to perform novel ­transformations beyond capabilities observed in a synthetic organic lab [1, 2]. Enabling Tools and Techniques for Organic Synthesis: A Practical Guide to Experimentation, Automation, and Computation, First Edition. Edited by Stephen G. Newman. © 2023 John Wiley & Sons, Inc. Published 2023 by John Wiley & Sons, Inc.

2

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

The attractiveness of using enzymes for transformations stems from their exquisite regio-­ and stereoselectivities  – something that traditional chemists still struggle to achieve in the lab  – that enzymes often execute effortlessly. Moreover, we have seen the emergence of multienzyme cascades for the synthesis of active pharmaceutical ingredients (APIs). A recent landmark example involves the synthesis of molnupiravir (MK-­4482), an orally dosed ribonucleoside analogue and inhibitor of influenza viruses, which has demonstrated activity against COVID-­19 when administered in animal models [3, 4]. In this work, McIntosh et al. developed a scalable three-­step route toward MK-­4482 [5]. Using a cascade of five enzymes, MK-­4482 could be accessed from 5-­isobutyrylribose (Figure 1.1). To the uninitiated, entering the world of enzyme-­catalyzed chemical transformations can be incredibly daunting, especially when one is not equipped with a foundational understanding of what an enzyme is and how these macromolecules work. However, you may be surprised to hear that enzymology and chemistry are not too different from each other at all! With an undergraduate chemistry background, a chemist can easily harness the power of enzymes to perform desired transformations – a fact that we aim to convince you of over the next few pages. However, while this chapter aims to illustrate the power of enzymes for novel and sustainable transformations, we do not want to inadvertently imply the use of these macromolecules is the be-­all-­end-­all solution  – sometimes the use of traditional organic synthesis to access target molecules is the more logical solution. Therefore, when an enzyme might be used is a weighted question often involving the combination of various intricate factors, including efficiency and cost. Over the following sections, we will do our best to educate you on these factors so that you can begin making an informed decision on this matter. We also aim to O NH N H HO

O HO

OH OH

NH

Enzymatic cascade

O

O

N

O

OH NH

NH2OH HMDS

O

N

O

O

O HO

OH

HO

OH

Molnupiravir (MK-4482)

O O

N

O

O

O

Figure 1.1  A combined enzymatic cascade/hydroxylamination for the synthesis of molnupiravir (MK-­4482).

O

1.1 ­Introductio

convince the reader that the use of enzymes is not limited to biologists and ­biochemists but also readily available for use by synthetic chemists. With the following breakdown of important considerations to make when using an enzyme, we hope to instill confidence in the reader that a biological catalyst is not too dissimilar to a chemical catalyst and can be readily obtainable from common suppliers. We will also dispel common misconceptions and myths surrounding the use of enzymes and then give an overview of several classes of reactions that can be performed with enzymes, including recent developments into more exotic transformations such as photobiocatalysis. This chapter will then conclude with a snippet into recent trends and technologies that have harnessed the use of enzymes in novel ways. We hope that the information gained from reading this chapter will provide a strong foundation for the reader to develop confidence in the use of enzymes and begin their venture into the world of biocatalysis.

1.1.2  Enzymatic and Organic Catalysis Are Not too Different from Each Other A seasoned chemist may be quite familiar with several stereoselective reactions whereby stereocontrol is dictated by the chiral environment of the reaction. For example, one model for the Corey–Bakshi–Shibata (CBS) reduction involves coordination of the respective carbonyl to the CBS catalyst in a specific spatial orientation, leading to stereoselective reduction of the carbonyl to the corresponding alcohol (Figure 1.2) [6]. Enzymes utilize a very similar concept to this reaction – the catalyst (enzyme) places a reactant (substrate) in a chiral environment (the active site), whereby stereoselectivity is dictated by the local reactive environment, leading to a selective reaction outcome. In the next subsection, we will look at how an enzyme achieves these feats.

Ph Me O RL

H RS

O B

+

Ph H

Ph

N

BH2

RL

–

O + O B– N H B H RS

Large group pseudoequatorial

Ph H

H Me

Ph

RL

Ph H

O – + O B N H BH2

RS

HO

H

RL

RS

Me

Figure 1.2  The CBS reaction has been used in undergraduate texts as a classic example of where an achiral reactant is stereoselectively transformed in a chiral environment to the corresponding product in high enantiomeric excesses.

3

4

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

1.1.3  Enzymes 101 Enzymes are known to accelerate reactions by more than 1017-­fold [7]. How an enzyme achieves these colossal rate increases under aqueous conditions requires an understanding of the active site architecture in great molecular detail. There are 20 essential amino acids found in nature. Enzymes are formed by cellular machinery, which stitch together combinations of these amino acids in a genetically pre-­defined sequence, making one very long polymer. This polymer is folded to give a precise three-­dimensional structure (Figure 1.3). The active site is defined as the region of the enzyme where substrates bind and undergo catalysis. The catalytic cycle begins with the binding of the substrate in the active site. This process precisely positions all molecules involved in the catalysis (metals, solvents, cofactors, etc.) in their respective orientations ready to achieve regio-­ and stereoselectivity. Subsequent activation of the substrate initiates the reaction, generating a transition state, which is stabilized by interactions with the active site residues of the enzyme. Following effective conversion of the substrate, the product is then released from the active site of the enzyme, completing one turnover and returning the catalyst back to its original state.

1.2  ­When Should I Choose an Enzyme over a Chemical Catalyst? The choice of using a chemical catalyst over a biochemical solution needs to be assessed on a case-­by-­case basis, often involving a detailed cost–benefit analysis. For example, chemical asymmetric imine reduction often requires the use of expensive precious metals, such as Ir, Rh, Ru, and Pd (Figure 1.4) [8]. While recent methods have moved toward Earth-­abundant solutions, such as employing iron or nickel, all these still require decoration with expensive chiral ligands that cannot be recycled [8], making the overall synthesis very environmentally and economically demanding. In contrast, imine reductases (IREDs) can perform stereoselective reductions without the use of expensive metals and can be performed under aqueous conditions mitigating the need for organic solvents. Since the initial report of IREDs in 2010  [9], many advancements have been made to use these enzymes for novel synthetic transformations [9, 10]. In fact, Matzel et al. published an elegant procedure for performing biocatalytic dynamic kinetic resolutions (DKRs) of aldehydes using IREDs (Figure  1.5). This method exploits the stereo-­preference of the enzyme for either the R-­or S-­chiral center [11]. The use of enzymes in this case showcases the re-­opening of the chemical window, enabling unprecedented reaction conditions, merging asymmetric reduction

HN

NH2 NH

5′-ATGTCTAGTAAATGCCTAGCTAGCTAGCT3′ 3′-TACAGATCATT TACGGATCGAT CGATCGA5′

Transcription Translation Protein folding

O

H N NH

N H S

DNA

O

O N H

NH O

O HN O

NH2 O

NH HN

NH O Protein

HO HN

HN

O

O

Figure 1.3  Proteins are formed by intracellular machinery that uses genetic information (DNA) to form a polymeric amino acid chain, which is then precisely folded to give a protein.

O

6

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis R3

N

R1

R3

Catalyst

R1

R2

Figure 1.4  Asymmetric reduction of imines to amines in the presence of a chiral catalyst.

NH R2

Imine formation R3 NH2

R1

O R2

R1

OH R2

Imine reduction R1

N R2

H2O

R3

IRED NADPH

Racemisation

NADP+

GDH R3 NH2

R1

O R2

H2O

R1

N

R1 R2

N H

R3

Cofactor regeneration

R3 Glucose Gluconic acid

R2

Figure 1.5  Biocatalytic dynamic kinetic resolutions of aldehydes using imine reductases. GDH = glucose dehydrogenase.

and water media. This would be near impossible to achieve with classical ­reducing agents, such as NaBH4 or Na(CH3COO)3BH. Next, we move the reader on to build an understanding of the variables associated with using a biocatalyst and here, they will gain foundational knowledge on how to gauge whether a chemical or biochemical solution is appropriate for ­solving a target problem.

1.3  ­Key Considerations for Running Biocatalytic Reactions With a primary level of appreciation of the power of enzymes, we can now continue our journey by addressing the variables that define a biocatalytic reaction. We will also illustrate how these variables change depending on the system that is being applied for the reaction. Before we advance in that direction, we will first begin by dissipating common myths surrounding the use of enzymes.

1.3.1  Dispelling Myths The uptake of biocatalysis in academic and industrial applications has increased significantly in recent years [12, 13]. Despite the positive perception of the technology, the breadth of applications remains rather modest. A factor that contributes to this lack of progress may be associated with the perceived notion of the limitations of biocatalysts – their availability, cost, ease of use, substrate scope, and operational stability. We aim to address these factors in the following subsections.

1.3  ­Key Considerations for Running Biocatalytic Reactions

1.3.1.1  Enzymes Are Not Safe to Use

Enzymes are very safe to use. In fact, proteins (as well as their essential metals and cofactors) are biodegradable and therefore considered to be environmentally friendly and can eliminate the need to use of noble metals (e.g. Pt, Pd, or Rh), which can complicate purification processes and are often subject to unforeseeable price variations. However, one must be careful not to inhale enzymes as they can illicit an immune response from susceptible individuals. Incidents such as these can be avoided by performing manipulations of reaction components under a fume hood wearing the appropriate protective wear. 1.3.1.2  Enzymes Are Not as Readily Available as Chemical Catalysts

One concern revolves around the availability of active biocatalysts. The acquisition of viable enzymes has come a long way. It used to be common for chemists to perform exhaustive screening of a broad variety of microorganisms, ranging from bacterial to fungal systems, to acquire a desired enzyme. Significant advancements to the methodology in which enzymes are manufactured has allowed them to be accessed in large quantities that may be readily available  [14]. Global efforts have allowed the modification and enhancement of enzymatic activities with site-­directed mutagenesis. This progress has been coupled with significant improvements in genomics and bioinformatics data, further allowing the rational and directional engineering of enzymes for unique activities. 1.3.1.3  Enzymes Are Seldom Useful Due to Their Limited Substrate Scope

Another misconception, which discourages the use of enzymes, is the myth that any given enzyme is only capable of performing a reaction with one specific substrate, whereby a different enzyme is required for each different substrate. While it is true that most enzymes have a strict substrate specificity, many can accept very different substrates and are not limited to their natural starting materials. Furthermore, significant advances in protein engineering techniques, such as directed evolution, have allowed biochemists to tailor the active site of an enzyme to accommodate non-­native substrates, allowing a broader range of novel biotransformations [12]. For example, Contente et al. demonstrated that a single point mutation in the active site of an acyl transferase in which a serine was substituted with a cysteine allowed the formation of thioesters, an observation that had not been previously reported with an acyltransferase [15]. Further examples include the work reported by Xu et al., where the conversion of an FAD-­dependent Baeyer–Villiger monoxygenase into a ketoreductase was achieved [16]. Moreover, Crawshaw et al. recently reported the engineering of a Morita–Baylis–Hillman-­ase to accept unnatural molecules as substrates, converting 2-­cyclohexen-­1-­one and 4-­nitrobenzaldehyde into 2-­(hydroxy(4-­nitrophenyl)methyl) cyclohex-­2-­en-­1-­one [17].

7

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

1.3.1.4  The Cost of Enzyme Production Is Very High

The use of biocatalysts in industry is often curbed by the myth of high production costs potentially associated with enzyme production. However, enzymes can often be purchased in different forms (see Section  1.3.2.7) each varying in price range. As depicted in Figure 1.6, producing 1 kg of unpurified enzyme ranges between €100 and €250, whereas the production of whole cells containing high levels of the desired enzyme is significantly cheaper at €35–€100 kg−1 [18]. It is important to also note that the price trend of purified enzyme formulations (e.g. FDH-­101 from Codexis or the immobilized lipase B) starts to skyrocket rapidly, approaching close to the prices of precious metals. However, this trend can also be seen in the case of metal catalysts. Moreover, even if the price of the metal used is insignificant, these often need to be further decorated with organic ligands which significantly increases the overall price. Case-­in-­point, Crabtree’s catalyst, which contains an iridium center, costs US$ 1 091 800 kg−1 of catalyst – the price of the adorned catalyst being six times higher than the metal alone. In addition, the prices of the metal cores are subjected to fluctuations depending on the stock market and their abundance in the earth’s crust. While the decision to choose a chemical or a biological solution may sound confusing, we can advise the reader to follow a simple checklist to assist in their choice of catalyst below. 1) Compare the performance of both chemical and biocatalyst reported in ­literature and choose the best option for the desired reaction. Cost comparison USD kg–1

Lipase B immobilized (2 U mg–1)

$1 28 048 $6 00 000

FDH-101 (Codexis)

$6 70 206

Rhodium $1 76 370

Iridium

Gold Ruthenium

$98 767 $69 490 $22 046

$10 91 800

00 10 00 0

00 80 00

00 60 00

00 00 40

00 00 20

0

Crabtree's catalyst

00

Palladium

00

Crude enzyme

$74 $192

12 0

Whole cells

Type of catalyst

8

USD kg–1

Figure 1.6  Cost comparison of chemical and biological catalysts in USD kg−1 of catalyst. Spot market @ 11 March 2022.

1.3  ­Key Considerations for Running Biocatalytic Reactions

2) Evaluate if either catalyst offers a significant benefit in, for example, selectivity, price, waste, pollution, etc. 3) If the chosen option is the biocatalyst, carefully read the literature to understand the physical form of the biocatalyst (see Section 1.3.2.7). In some cases, use of the enzyme in one or another form is preferred, for example, often when using whole cells the product accumulates inside the cell, or when using crude extracts one can have cross reactivity with the cell metabolites. 4) The most common choice for preferred enzyme form are purified enzymes. However, in some cases, the use of this enzyme form can be excessive since the reaction might also be efficiently performed with crude extracts or whole cells, thus reducing significantly the cost of the catalyst. 1.3.1.5  Enzymes Are Functionally Unstable Under Organic Conditions

The functional stability of biocatalysts is one of the principal parameters to address in the design of a viable enzyme and has steadily improved over the past few years. Such developments are crucial, especially if the reaction conditions stray from aqueous conditions, such as in the use of high ratios of organic co-­ solvents. Genetic engineering has allowed the design of highly active enzymes which operate under conditions of high temperatures, non-­buffer salts, solid-­state buffers, crown ethers, and cyclodextrins [19]. Furthermore, several remedies can be taken to boost the performance and stability of the biocatalyst, such as entrapment or immobilization onto a solid support (see Section 1.5.2). 1.3.1.6  Sustainability

Biocatalysis has been described many times as the Holy grail and perfect solution for reducing the waste and contaminants produced in industry – but is this always the case? The answer is definitely no. Even if most of the biocatalytic transformations are more environmentally friendly than classical chemical synthesis, sometimes it is better to stick to the chemical process. For example, we would not recommend that an enzymatic process is used to access multi-­kilo quantities of product since some biocatalytic processes are limited to very low substrate concentrations (μm–mm). However, while making this point, we truly believe that the future of chemical manufacturing will make chemoenzymatic approaches feasible, where the advantages of organic synthesis and biocatalysis can be merged [20].

1.3.2  Challenges of Using Enzymes: the Need for Strict Reaction Conditions Now that we have addressed some common myths, let us look at important considerations needed when addressing an enzymatic reaction. While procedures for

9

10

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

biocatalytic reactions are well documented, the identical replication of a published procedure gives no guarantee of a similar result. This observation is often due to the multitude of factors that contribute to the biocatalytic transformation in question, including that of pH, ionic strength, and temperature of the reaction media, to name a few. Moreover, we should not lose sight that enzymes are naturally evolved biomacromolecules coming from, and performing reactions inside, living cells and we, as chemists, are essentially “hijacking” this natural process to carry out novel synthetic transformations. Therefore, we need to make careful considerations to ensure the enzyme is fully functional to perform catalysis in the chemistry lab. Here, we outline a few key conditions to be aware of when designing your biocatalytic assay. 1.3.2.1  Enzymes from Extremophiles

An understanding on the bacterial origins of any enzyme is key in exploiting the full potential of its reactivity. For example, enzymes from thermophilic organisms are endowed with structure–function properties of high thermostability and optimal activity at these high temperature ranges while having poor functionality at moderate temperatures below 40 °C [21]. Psychrophiles, on the other hand, produce enzymes that may only function at cold temperatures (15 °C or lower). Halophiles thrive in environments of high salt concentrations and therefore their enzymes may require such conditions when used in biocatalytic applications. Finally, acidophiles grow in environments of high acidity, at pH 2.0 or below. 1.3.2.2  Solvents (and Co-­solvents)

A hallmark feature on the use of enzymes is the ability to perform reactions in water. However, in some cases, the aid of a co-­solvent, such as ethanol, acetone, DMSO, or acetonitrile, is necessary. Not only does the addition of a co-­solvent assist in the solubility of nonpolar substrates under aqueous conditions, it can also enhance the activity and stability of enzymes, aid in product recovery, or even shift the equilibrium toward new reactions [22–24]. With significant advances made in the understanding of protein structure and dynamics in the past few decades, there are many reports on the engineering of proteins for nonaqueous biocatalysis [25]. Furthermore, with the significant leaps made in computational methodologies, more options are becoming available for the prediction and design of enzymes, which work better under a set of reaction conditions. For example, Cui et al. recently published a guide on using molecular dynamics (MD) simulations to assist in the rational design of proteins under organic co-­solvent conditions [26]. 1.3.2.3  Concentration and Ionic Strength of the Buffer

Enzymes typically only function in aqueous environments at a specific pH, the latter often being dictated by the natural environment of the organism from which

1.3  ­Key Considerations for Running Biocatalytic Reactions

the enzyme originated. Consequently, buffers are used to adjust and stabilize the pH of the solution during an enzyme reaction. While the design of buffer systems is beyond the scope of this chapter, the ionic strength and concentration, as well as the choice of buffer components will be addressed here. Salt concentration can have dramatic effects on the binding and turnover of a substrate by an enzyme. Moreover, the choice of ion (i.e. Na+ vs. K+, Cl− vs. Br) can have a significant effect on the ionic strength of the buffer, as well as enzymatic activity. Case-­in-­point are enzymes that require either a mono-­or divalent cation for catalysis [27]. Supplementation of the buffer with the wrong ion can lead to a  slow or inactive enzyme. Some buffers, for example those containing TRIS (tris(hydroxymethyl)aminomethane) or even the commonly used phosphate buffer, can also have destabilizing effects on protein structure [28]. Therefore, it may be useful to dedicate a small amount of time investigating previously reported conditions for a particular enzyme in the literature to determine what salts are compatible. The buffer concentration dictates the capacity of a buffer to stabilize the pH, with higher concentrations having higher buffering capacity. This is typically 0.05–0.2 M for most enzymes; however, halophilic and thermophilic enzymes are known to prefer higher concentrations of up to 1  M. Effects of substrates and additives (whether these are sold in their salt forms) on the ionic strength of the buffer must also be considered. 1.3.2.4  pH Dependence

The activity of an enzyme is strictly dependent on the pH of the reaction, which can be explained by a couple of factors; first, the state of protonation of the functional groups of amino acids and cofactors involved in the reaction will invariably change with change in acidity. This effect could have drastic consequences to catalytic activity, especially if the residues directly involved in catalysis are dependent on specific protonation states. While the pH optima of many enzymes are found near physiological pH, there are some that deviate from this rule – for example, the stomach protease, pepsin, works with a pH optimum of 2, while alkaline phosphatase has a pH optimum of 10.5. Second, variations in pH can lead to irreversible damage to the three-­dimensional structure of the protein. Therefore, assays of an enzyme to find the pH optima may be necessary and significant deviations from the pH optimum should be avoided. 1.3.2.5  Concentration of Reactants

Reactant concentration is a key parameter that needs to be carefully analyzed while planning a biocatalytic reaction. The first obstacle that one can encounter is the solubility of the substrate; many organic compounds are not soluble at high concentrations in the reaction media (generally water). To tackle this issue, ­co-­solvents can be added.

11

12

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

One also needs to keep in mind that the inherent nature of enzymes is to work inside cells where the substrate concentrations are generally very low compared to the protein [29]. Therefore, increasing the substrate concentration may lead to initial saturation of the enzyme activity, followed by subsequent inhibition. 1.3.2.6  Enzyme Concentration

In a manner similar to using a chemical catalyst, the amount of enzyme to add to a reaction is a balance between achieving the highest reaction conversion in an appropriate amount of time – this differs from enzyme to enzyme. However, if an enzyme is commercially available, then it is likely that a procedure is already available. In such a case, one just needs to emulate the published reaction conditions using the reported values as a starting point for determining the amount of enzyme to add. 1.3.2.7  Enzyme Forms

Enzymes can be stored/sold in purified form (only the desired enzyme is present) or as a crude extract (where the desired enzyme is present along with other proteins and cell metabolites released when the enzymes are extracted out of the cell). More commonly, enzymes are stored in solution since they come from microorganisms and chemistry in nature is mostly performed in solution. However, in the lab they can be prepared/formulated in different ways depending on the desired purpose (Figure 1.7). Whole cells – most enzymes are first synthesized within a host cell using the RNA/DNA machinery and then lyophilized to give lyophilized cells containing the desired enzyme. These cells can also be engineered to produce particularly large amounts of a given enzyme and may be harvested and used directly for catalysis. This method is useful when the biocatalyst is unstable under extracellular conditions. From an economical point of view, this form is very attractive ­considering that isolation and purification of the biocatalyst is avoided. Enzyme forms Whole cells

Solubilized

Lyophilized

Immobilized

Entrapment

Lyophyilized cells containing SHC enzyme

Solution of MsAcT (wt) in HPO4 buffer pH: 8.0

Lyophyilized MsAcT (wt)

Glass column packed with immobilized MsAcT (wt) onto glyoxyl agarose beads.

Entrapped SHC enzyme into polyacrylamide

Figure 1.7  Forms that enzymes can be stored/sold in.

1.3  ­Key Considerations for Running Biocatalytic Reactions

Solubilized – the most common way of using enzymes in vitro involves storing them in an appropriate buffer. In this case, no additional treatment of the enzyme is required after purification (cf. whole cells or lyophilized enzyme, which need rehydration). Lyophilized – this technique is often employed when long-­term storage of the enzyme is required. However, it is important to note that some enzymes may suffer from deactivation since water may be required to stabilize their quaternary structures. This method can be applied to either purified or crude extracts. Immobilized – immobilization of an enzyme onto a solid support is particularly advantageous when enzyme recovery and re-­use is desired, i.e. flow biocatalysis. This method is applicable to purified enzymes and whole cells. Entrapment – entrapment of an enzyme into a polymeric matrix is a milder immobilization technique, since in this case, the enzyme does not suffer conformational changes due to chemical bonding to the support. 1.3.2.8  Toxicity

When addressing the toxicity of enzymes, one must distinguish between pure enzymes and whole cell biocatalysts. While, to the best of our knowledge, pure enzymes are not toxic (aside from some reported hypersensitivity cases in enzyme replacement therapies [30]), using whole cells can pose risks due to the potential toxicity of cell metabolites. Even the cell wall fragments can present pathogenic roles  [31]. In addition, the recombinant DNA of these type of cells encodes an antibiotic resistance gene that can be transformed or eventually conjugated between bacteria, leading to populations of antibiotic-­resistant bacteria [32]. No special considerations need to be taken into account when working with regular Escherichia coli cells (most of the enzymes are synthesized in E. coli), apart from the standard personal protective equipment used in a regular chemistry lab (lab coat, safety goggles, and gloves). Even if some variants of E. coli are completely innocuous, as are the species commonly found in the large intestine of humans [33], others may contain virulence factors. Therefore, one needs to treat this situation similar to organic synthesis, whereby a newly synthesized compound is treated with utmost care due to no available safety data for that molecule.

1.3.3  What Do I Need to Start Biocatalytic Experiments in My Lab? The running of a biocatalytic experiment in a chemistry lab does not require many unfamiliar pieces of equipment (Figure 1.8), with perhaps the most exotic piece of equipment needed being a spectrophotometer (not compulsory but highly recommended, see Section 1.6). In a similar manner to an organic wet lab, a protocol needs to be written down, conditions thoroughly considered, and reagents and

13

14

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

Equipment list 1. Set of micro-pipettes (1–20, 20–200, 100–1000) μL 2. Round-bottomed flasks 3. Bench-top spectrophotometer 4. Set of centrifuge tubes (0.5, 1.5 and 2.0) mL 5. TLC plates 6. Hot plate magnetic stirrer

Figure 1.8  Standard pieces of equipment needed for running a biocatalytic reaction can be found in most organic chemistry wet lab setups.

equipment pre-­prepared before mixing. We point the reader toward a detailed ­outline by Bisswanger  [34], where practical considerations when starting a ­biocatalytic reaction are neatly summarized.

1.3.4  Additional Considerations Upon storage, enzymes can suffer from aggregation, precipitation, and other structural changes that may affect their activity. This parameter is key in every enzymatic reaction; it not only tells us about the performance of the enzyme towards a specific reaction, but also sets a baseline for the maximal amount of activity expected from that specific batch of the enzyme over time. Let us observe the following example: a chemist is attempting a transesterification with the commercially available lipase, CalB, that comes lyophilized. Prior to setting up the reaction, the chemist makes a stock solution of 1 mg mL−1 enzyme in phosphate buffer at pH 8.0. After running the first set of reactions, they obtain an average of 90% conversion. Repetition of same transformations by the same chemist after two weeks with the same stock solution still furnishes the same product, but surprisingly, no greater than 40% conversion can be obtained. How is it possible to obtain such a dissimilar result? Did the chemist do something wrong? The answer is: probably not. Many enzymes are not stable over time, meaning that they gradually lose activity. Therefore, as a rule in biocatalysis, prior to each reaction, one must check the initial activity of the enzyme toward a known substrate. In this way, a baseline level of activity is set and, if now after a certain time the chemist wants to retry the reaction with the same enzyme, can have an idea of how much activity was lost during storage. The specific activity of an enzyme can be garnered by the enzyme unit (U), where 1 U is defined as the amount of enzyme required to convert 1 μmol of substrate per minute under the specified conditions  [35]. If purchased commercially, the amount of enzymes which constitutes to 1 U is provided on the commercial bottle in units of U mg−1;

1.4 ­Transformations Catalyzed by Enzyme

however, if you have a non-­commercially available or non-­reported enzyme, you may need to create an assay that allows you to define 1 U for that particular enzyme. The use of the unit allows a quantification of the amount of an enzyme with respect to its function rather than its mass. While the use of the enzyme unit is not presently valid according to the SI system (the unit was replaced with the katal in 1978  [36]), most commercial suppliers of enzymes continue to market enzyme preparations in enzyme units. As in any other chemical transformation, a reaction work-­up is usually necessary in order to separate the products from the biocatalyst. To do so, liquid–liquid extractions in the case of soluble enzymes are generally effective. In the case of supported/immobilized enzymes, an additional filtration may be required.

1.4  ­Transformations Catalyzed by Enzymes In this section, we will glance over how enzymes are categorized into various classes and provide some examples of commercially available enzymes that have been used to perform biotransformations in standard chemistry laboratories. This section aims to provide the reader of a scope of possible reaction types they may want to conduct.

1.4.1  EC – The Enzyme Commission Number The classification of enzymes follows the Enzyme Commission (EC) number, which categorizes each protein according to the type of reaction the enzyme catalyzes. An EC number is composed of four digits in the form, EC x.x.x.x. The first number denotes the class of the enzyme. For example, one class of enzymes is called “oxoreductases.” All oxoreductases will have an EC classification of the format, EC 1.x.x.x. The second number gives the subclass, with the third number further dividing the subclass into the sub-­subclass. Finally, the fourth number represents the serial number of the enzyme in its sub-­subclass. As of 2022, there are seven EC levels: the oxoreductases, transferases, hydrolases, lyases, isomerases, ligases, and translocases (Figure 1.9). 1.4.1.1  EC 1 – Oxoreductases

Oxoreductases (also known as oxireductases) catalyze the transfer of electrons from an electron donor (reductant) to an electron acceptor (oxidant), often using ­cofactors (flavin, heme, and other metal ions). One important class of ­oxoreductases – the dehydrogenases – catalyze the oxidation of primary or secondary alcohols to the respective carbonyl product (Figure 1.10) [37]. Oxoreductases make up one-­third of all proteins classified in the BRENDA database [38].

15

16

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis Dehydrogenases Oxidases Reductases

ABC transporters

Transaldolases Methyltransferases Transaminases

EC 1 oxoreductases

EC 2 transferases

EC 7 translocases

Enzyme classification Synthases Carboxylases Ligases

Glycosylases Peptidases Nucleases

EC 3 hydrolases

EC 6 ligases

EC 5 isomerases

EC 4 lyases

Racemases Epimerases

Decarboxylases Aldolases

Figure 1.9  Classification of enzymes based on reaction type with several examples from each class.

HO R

H H′ R

O NH2

N

R

H′ O

H

O

ADH

NH2

R

R

N R

Figure 1.10  Generic reaction catalyzed by alcohol dehydrogenase.

1.4.1.2  EC 2 – Transferases

This class of enzymes catalyze the transfer of a specific group from one molecule to another. Groups that can be transferred include, but are not limited to, methyl, acyl, amino, glycosyl, and phospho-­groups. For example, aspartate aminotransferase (AspAT) catalyzes the transfer of an amino group from l-­aspartate to 2-­oxoglutarate to give l-­glutamate and oxalacetate (Figure 1.11) [39]. 1.4.1.3  EC 3 – Hydrolases

Hydrolases have the capacity to catalyze the hydrolysis of various substrates. Depending on their substrate specificity, the hydrolases are further divided into

1.4 ­Transformations Catalyzed by Enzyme O HO

O –

O

O

O

O

–

AspAT

–

O

NH 2 O L-aspartate

O

O

–

O

O

–

–

O NH 2 L-glutamate

O oxalacetate

O 2-oxoglutarate

O

HO

Figure 1.11  Interconversion mediated by aspartate aminotransferase.

O

OH OH

OH

OH

N

CAL-B, 30 °C organic solvent

F

NC

NC

O N

O

NC

O

NC

F

OH +

F

(R)-(+)

N

MsCl/Et3N

O

N

F (S)-(+)-Citalopram

(S)-(–)

Figure 1.12  Chemoenzymatic synthesis of (S )-­(+)-­Citalopram.

their respective subclasses. One class of hydrolases, the lipases, have been particularly useful in the synthesis of novel pharmaceuticals. For example, an enzymatic acylation of racemic 4-­[4-­dimethylamino]-­1-­(4′-­fluorophenyl)-­1-­hydrox-­1-­butyl]-­ 3-­(hydroxymethyl)-­benzonitrile using CAL-­B and vinyl acetate as the acyl donor allowing the synthesis of (S)-­(+)-­Citalopram (Figure  1.12)  [40]. The hydrolases are further subclassified into – EC 3.1 ester hydrolases; EC 3.2 glycosylases; EC 3.3 enzymes hydrolyze ether linkages; EC 3.4 peptidase/protease, etc. 1.4.1.4  EC 4 – Lyases

Lyases catalyze non-­hydrolytic bond cleavage reactions, ranging from C–C, C–O, C–N, and C–S, leading to the formation of a double bond, a new ring, or addition of groups to double bonds. The lyases are consequently subdivided into the bond type involved in the reaction: EC 4.1 carbon–carbon lyases, EC 4.2 carbon–oxygen lyases, EC 4.3 carbon–nitrogen lyases, EC 4.4 carbon–sulfur lyases, EC 4.5 carbon–halide lyases, EC 4.6 phosphorus–oxygen lyases, EC 4.7 carbon–phosphorus lyases, and EC 4.99 other lyases. Lyases are important in many biological processes – for example, as the first committed step towards the phenyl propanoid pathway, phenylalanine ammonia lyase (PAL, EC 4.3.1.24) catalyzes the conversion of l-­phenylalanine to ammonia and trans-­cinnamic acid (Figure 1.13) [41]. Figure 1.13  Reaction catalyzed by phenylalanine ammonia lyase (PAL).

O

O OH

NH2

PAL

NH3

+

OH

17

18

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

1.4.1.5  EC 5 – Isomerases

Isomerases catalyze intramolecular rearrangement or isomerization reactions. This class of enzymes is subdivided into seven subclasses based on the type of catalyzed reaction. EC 5.1 racemases and epimerases, EC 5.2 cis–trans isomerases, EC 5.3 intramolecular oxidoreductases, EC 5.4 intramolecular transferases, EC 5.5 intramolecular lyases, EC 5.6 isomerases altering macromolecular conformation, and EC 5.99 other isomerases. One classic example from undergraduate biochemistry classes is glucose-­6-­phosphate isomerase  [42], which functions in glycolysis by interconverting glucose-­6-­phosphate and fructose-­6-­ phosphate, the latter which proceeds through the glycolytic pathway to generate ATP (Figure 1.14). 1.4.1.6  EC 6 – Ligases

Ligases, as implied by their name, catalyze the attachment of two molecules. For example, the enzyme pyruvate carboxylase (EC 6.4.1.1) catalyzes the carboxylation of pyruvate to form oxaloacetate (Figure 1.15) [43]. Reactions catalyzed by ligases require energy from nucleoside triphosphates [44]. 1.4.1.7  EC 7 – Translocases

Translocases assist in the movement of ions or molecules across the cell, often involving the hydrolysis of ATP. The translocases can be further subdivided into six subclasses, depending on the type of ion or molecule being translocated. EC 7.1 enzymes catalyze the translocation of hydrons, EC 7.2. inorganic cations, EC 7.3. inorganic anions, EC.7.4. amino acids and peptides, EC 7.5. carbohydrates, and EC 7.6. other compounds.

O

2–

OPO3

OH

Phosphoglucose isomerase

HO

OH OH D-Glucose-6-phosphate

2–

OPO3

O CH2OH HO OH OH α- D-Fructose-6-phosphate

Figure 1.14  Reaction catalyzed by glucose-­6-­phosphate isomerase.

O OH O

–

+ HCO3 + ATP

Pyruvate carboxylase

O HO

O OH

O

Figure 1.15  Reaction catalyzed by pyruvate carboxylase.

+ ADP + Pi

1.4 ­Transformations Catalyzed by Enzyme

1.4.2  Some Applications of Selected Commercially Available Enzymes 1.4.2.1  Horseradish Peroxidase

Horseradish peroxidase (HRP) is an enzyme that is found in the roots of horseradish and is commonly used to catalyze the oxidation of various organic substrates using hydrogen peroxide. In biochemistry, HRP is used to amplify signals in photometric assays, whereby chromogenic or chemiluminescent substrates are oxidized for the detection of targets, such as proteins and nucleic acids  [45]. For example, in chemiluminescent Western blotting, a technique that enables indirect detection of protein samples immobilized on membranes, a primary antibody is applied against the protein of interest. After subsequent washing of excess primary antibody, a secondary antibody, which targets the primary antibody and is conjugated with HRP, is added, before another washing procedure is applied. Following this, a buffer containing peroxide and luminol is added, in which the latter oxidizes and forms an excited state product that emits light as it decays to the ground state (Figure 1.16). 1.4.2.2  Lysozyme

Lysozyme, also known as muramidase, is a glycosyl hydrolase that catalyzes the hydrolysis of 1,4-­β-­linkages between N-­acetylmuramic acid and N-­acetyl­d-­glucosamine residues in peptidoglycan, a major component of gram-­positive bacterial cell walls (Figure 1.17). This reactivity consequently disrupts the integrity of the cell wall, causing osmotic shock and leading to lysis of the bacteria. Commercially available hen egg white lysozyme is commonly used for the aforementioned purpose; however, this enzyme has also enjoyed the spotlight in ­several biotechnological and catalytic applications [46, 47].

Luminol

HRP Primary antibody Secondary antibody conjugated with HRP Protein of interest

Figure 1.16  Use of horseradish peroxidase in Western blotting.

19

20

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

RO

OH O O HO O NH

O

R

O O

NH O

O OH R

O

OH O O HO O NH

HO O

Site of hydrolysis OH O O HO O NH

O O

NH O

OR Lysozyme H2O

OH

O O

R

HO

O

R

O O

NH OH

O OH

OH O O HO O NH

NH O

O

OH

Figure 1.17  Reaction catalyzed by lysozyme.

1.4.2.3  Trypsin

Trypsin is a serine protease found in the digestive tract of many vertebrates. This enzyme catalyzes the hydrolysis of proteins, cleaving the carboxyl end of lysine or arginine residues. As such, trypsin has been very useful in proteomics where it has been used to digest proteins for mass spectrometric analysis [48]. 1.4.2.4  Candida Lipase B

Candida antarctica: lipase B (CALB) is a hydrolase that has been used in a broad range of organic reactions, including kinetic resolutions [49], aminolysis [50], and even reactions in organic solvents [51]. Having been extensively studied, CALB has undergone several modifications via mutagenesis, for example, to improve its thermal stability [52] and activity under immobilized conditions [53, 54]. A commercially available immobilized version of CALB, Novozyme® 435, has been used for the stereoselective hydrolysis of esters, transesterification, and DKR of alcohols [55, 56]. 1.4.2.5  Amino Acid Dehydrogenase

Amino acid dehydrogenase enzymes employ NAD+ and H2O to catalyze the deamination of amino acids (Figure 1.18) [57]. Depending on the specificity of the amino acid dehydrogenase, either l-­ or d-­amino acids can be converted to their corresponding oxo-­acid. d-­Amino acid dehydrogenase has been particularly useful in the synthesis of d-­amino acids in both high enantioselectivity and high yields [58]. R

R OH + H O + Acceptor 2

H2N O

OH + NH + Reduced acceptor 3

O O

Figure 1.18  Transformation catalyzed by d-­amino acid dehydrogenase. Acceptor can be various electron acceptors, but commonly Coenzyme Q is used.

1.5  ­New Trends and Technologies in Biocatalysi

1.4.2.6 Glycosidases

The inventory of commercially available enzymes which catalyze the hydrolysis of sugars is vast, largely due to the strict specificity of each enzyme for their respective glycosidic linkage. Glycosidases have been particularly useful in biotechnology where they have been used for various purposes, including improving the quality of bakery products, brewing beer, and biomodification of materials [59].

1.4.3  Engineered (Unnatural) Reactions As our understanding of enzymes at the atomic level improves, the development of powerful bioinformatic tools and machine learning algorithms has also flourished. Consequently, these advancements have allowed the development of “designer” enzymes that catalyze novel reactions [16]. Tools such as AlphaFold, an AI system developed by DeepMind, has the power to predict protein structure based on genomic datasets [60]. RetroBioCat, a tool for computer-­aided design of biocatalytic cascades, also allows for easy searching of a set of enzymes that might perform a desired transformation [61]. We refer to recent advancements in these fields in Section 1.5.

1.5  ­New Trends and Technologies in Biocatalysis 1.5.1  Flow Biocatalysis and New Technologies 1.5.1.1  What Is Flow Biocatalysis?

In the last few decades, flow chemistry has risen as a promising alternative to perform chemical synthesis in a more efficient, selective, sustainable, and safer manner (see Chapter 4) [62–64]. Flow biocatalysis has allowed a merger between continuous processing and enzyme-­mediated reactions, forming the perfect marriage for sustainable and efficient biocatalytic transformations [65–67]. As with any continuous processing setup, the reagents are pumped through a system by means of pumps through modules. Implementation of flow biocatalysis begs several questions – how does this process work? How does one decide when a flow compared to a batch process is beneficial for a specific transformation? Can every enzymatic reaction be implemented in flow? In the following paragraphs, we aim to guide you on when to consider a flow setup. 1.5.1.2  How Does Flow Biocatalysis Work?

The equipment required for flow biocatalysis does not far exceed a generic flow setup one may already have set up in a chemistry lab so can be easily implemented into an existing framework. Due to its modular nature, flow systems can be viewed

21

Reagent A

LiquidLiquid extractor

Pump A

Tubing

Purification column

Distillation unit

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

Reagent B

22

Reactor Mixer

Pump B

Tubing

Figure 1.19  Representation of the typical units and general setup for flow chemistry experiments.

in a similar manner as “Lego” – the user can change the bricks (modules) depending on their specific needs (Figure 1.19). The different “Lego” bricks can represent the analytical, purification, solvent switchers, reactors, and an infinite number of other units. However, where does the enzyme take part in this scientific Lego? For optimal sustainability, the enzyme needs to be recycled and reused after each synthesis. Recently, a plethora of techniques to immobilize the enzyme inside the reactor have been reported [68–70], enabling a continuous flow of reagents without losing the precious catalyst [71]. Different methods of immobilization can be found throughout the literature and generally classified into chemical (covalent) and physical methods (embedding) (see Figure 1.20). 1) Chemical immobilization of enzymes refers to the covalent binding of the enzyme onto a solid support. This technique offers several advantages, such as stabilization of the biocatalyst due to rigidification [72] and low leaching after multiple catalytic cycles. 2) Physical methods involve the embedding of an enzyme into an inorganic or organic polymer. Physical entrapment of the enzyme is a milder procedure compared to chemical immobilization since the enzyme is not stressed to undergo conformational changes due to the chemical linkage to the solid supports [73]. An innovative example of physical immobilization is 3D bioprinting, where an enzyme is extruded together with a polymer into a desired shape, opening the door to an infinite type of micro-­and meso-­reactors [74]. This method can prove advantageous when dealing with enzymes that suffer inactivation upon

1.5  ­New Trends and Technologies in Biocatalysi

Adsorption/ ionic

Chelation

Packed bed reactors L

O O N

O O

Chemical

Co+2

L N

NH

N HN

Covalent

S

S

S S

Immobilization methods and applications

Cross-linking 3D bioprinting Physical

Entrapment

Figure 1.20  General enzyme immobilization techniques and their potential applications.

covalent or ionic immobilization due to conformational changes in their quaternary ­structure [75, 76]. 1.5.1.3  When Is a Flow Process More Beneficial for a Specific Transformation?

Several considerations need to be made when deciding whether to switch from a classical batch process to a flow setup (applies in both classical synthesis and biocatalysis). One of these is process safety – flow setups offer safety levels that are astonishingly higher than batch for handling dangerous or very reactive intermediates due to enhanced heat transfer. Even though the reactions are usually performed under milder conditions in biocatalysis, more harsher conditions may be introduced downstream if running a chemo-­enzymatic synthesis. Another common issue of batch reactions involves product build-­up, leading to the inhibition of the enzyme, and therefore decreasing the productivity of the reaction. In the case of a flow setup, the products are continuously evacuated from the reactor where the enzyme is confined, mitigating this problem [67]. Moreover, when a reaction scale-­up is necessary, a flow setup can prove very beneficial since the process follows a more linear trend, and reaction parameters can be screened much faster than in batch mode. 1.5.1.4  Should One Implement Every Enzymatic Reaction in Flow?

Flow setups can have weak points (long reaction times, insoluble compounds, etc.), and in some cases, a batch reaction is still the best option. The typical approach for deciding if a flow setup could be beneficial for a specific reaction would be first running the reaction initially in batch to identify the critical

23

24

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

parameters (such as reaction times, solubilities, temperatures, etc.) and then ­moving the reaction to flow to determine if any improvement is observed.

1.5.2  Enzyme Engineering Generally, enzymes exhibit different activities toward different substrates. To improve the enzymatic activity toward a specific substrate or class of molecules, the structure of the enzyme can be finely tuned using enzyme engineering, which, in classical organic chemistry would be akin to changing the ligands around the metal center in order to tune the selectivity of the catalyst. This can lead to the improvement of different parameters, such as temperature, pH, solvent tolerance, substrate, and reagent specificity, among others. There are several methods to engineer new enzyme variants (mutants), which are summarized in Figure 1.21. The model-­driven approach is the most classical one in which a mutation (changing a ligand by another one in organometallic chemistry) (or a set of mutations) can be planned by taking into consideration the physicochemical properties and interactions of the different residues of the protein. Complementary to the model-­driven approaches in the last decades, data-­driven approaches are proving to be very fruitful [77]. The main difference between both methodologies is that the latter is based on finding patterns and trends in pre-­existing data to predict properties (i.e. solubility,

Improved enzyme

Data driven

Model driven

Protein engineering Rational design

Machine learning

Semi-rational design

Artificial intelligence

Directed evolution

De novo design

Figure 1.21  Two approaches of enzyme engineering.

1.6  ­Flow Chart to Biocatalysi

thermal stability, activity). Machine learning and artificial intelligence have been demonstrated as excellent tools for analyzing multidimensional relations that the human brain is not able to identify [78]. In between model-­ and data-­ driven approaches, de novo design is emerging as a promising alternative to fabricate new proteins from scratch, based on experimental data and computer-­ aided design [79, 80].

1.5.3  Photobiocatalysis Photobiocatalysis has recently gained a significant amount of interest by the scientific community [81]. The biocatalyst can act either as a chemical environment to favor the transformation of the photochemically activated substrate [82] or as mediator in which the enzyme converts light energy into redox equivalents [83]. Not surprisingly, in the last few years, photobiocatalysis has been merged with continuous flow technologies [84, 85].

1.6  ­Flow Chart to Biocatalysis To simplify the task of choosing the right enzyme, we have developed the following flow chart in which we guide the reader through the crucial steps when dealing with a biocatalytic reaction (Figure 1.22). 1) Identify the target reaction First, we need to determine the transformation we want to achieve so that we may ascertain the enzyme required. These transformations may be oxidation, reduction, C–C bond formation, C–C bond cleavage, transamination, reductive amination, etc. Hint: Many enzymes are named after the reaction they catalyze. 2) Is there a biocatalytic route available? The chemist now needs to explore the literature to identify an enzyme or a class of enzymes that can perform the desired transformation (SciFinder, Web of Science, Reaxys, etc.). If one is available, you can go on to step 3. However, if the desired enzymatic transformation has not been reported, it is unlikely to be easily achieved and beyond the scope of the expertise of an average organic chemist. 3) Is there a commercial enzyme available? When looking for commercial enzymes, we suggest browsing through: Sigma Aldrich, Gecco, Codexis, Creative-­enzymes, Eucodis, Purolite, Resindion, and Novozymes. These sources have a large repertoire of enzymes neatly catalogued for perusal. If the enzyme is commercially available, the chances of success are much higher, since it is very likely the protein has been thoroughly

25

26

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

1) Identify the target reaction

2) Is there a biocatalytic route available?

No

Traditional synthesis is still the best option

No

3′) Contact a biochemistry group or related

Yes

3) Is there a commercial enzyme available?

Yes

4) Choose the right enzyme depending on your needs: • Temperature • Solvent • Form: soluble or immobilized • Type of support if immobilized (hydrophilic/lipophilic)

5) Perform the standard activity assay

6) Test your own substrates

Figure 1.22  Flow chart to biocatalysis.

characterized. Often, many of those suppliers will offer screening kits to rapidly find a hit. However, if the biocatalyst is not commercially available, it is still possible to engineer an enzyme for the target transformation, but this process is very complex and will likely require a collaboration with a specialized biochemistry group. 4) Choosing an enzyme based on your needs Once we know that a commercial supplier or a biochemistry group can provide us with the biocatalyst, we need to consider the compatibility of the enzyme with the reaction conditions. For example, if the substrate is only soluble in a specific organic solvent, the enzyme must be compatible/stable in it. Moreover, one needs to decide whether enzyme recovery is required at the end of the reaction. For example, if your enzyme is difficult to make in large quantities,

1.7  Case Study: Setting up a Biotransformation

requiring multiple rounds of expression and purification, then in such a case, an immobilized enzyme may be the best option, allowing you to retain the enzyme for multiple rounds of catalysis before disposal (see Section  1.5.2). Depending on the polarity of the substrate, a lipo-­ or hydrophilic support may prove advantageous. Contrastingly, the use of very apolar substrates would not be suitable for such a support. Vendors such as Sigma Aldrich, Purolite, and the other previously reported suppliers have different supports and immobilization chemistries available. We invite the reader to consult the respective catalogues to be informed of what options are available. 5) Perform a standard activity assay Once we have the enzyme at hand, a standard activity assay must be performed to assure the healthiness of the enzyme, as discussed in Section 1.3.3. Ideally but not compulsory (NMR, HPLC, GC can also be used), this assay will be carried out in a classical spectrophotometer since most of the reported activity assays are a measure of the change in absorbance of a certain molecule therefore, they are usually very fast. When using commercial or published enzymes, the activity assays should be thoroughly described in the Supplementary Information section of the respective publication. If the recovered activity fits the reported values, one can move on to the last step of the process. 6) Test your own substrates Finally, screening of the desired substrates can be performed. As a rule of thumb first, the published reaction conditions will be tested (enzyme concentrations, solvents or co-­solvents, temperature, reagents concentrations, etc.) and then adjusted to the specific starting materials if needed. Regarding the workup of the reactions, usually a regular reaction workup should enable product recovery, nevertheless when using immobilized enzymes an additional filtering step should be performed. The reactions can be analyzed with the usual techniques such as NMR, HPLC, or GC. No special techniques or skills should be needed for an organic chemist to perform biocatalytic reactions.

1.7  Case Study: Setting up a Biotransformation In this section, we aim to guide the reader through a case study of the process of setting up a biocatalytic reaction, from identifying the transformation in the literature until the final reaction setup, using the flow chart depicted in Figure 1.22. We have chosen the “Enzymatic Kinetic Resolution of (R/S )-­1-­phenylethanol” [86] as an elementary example for the reader to follow. 1) Identify the target reaction The reaction we are interested in here is the kinetic resolution of a ­racemic mixture of 1-­phenylethanol (Figure 1.23).

27

28

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis O OH

OH

Lipase, acyl donor

O

Organic solvent 35 °C, 1.5 h (R/S)

(S)

(R)

Figure 1.23  Enzymatic kinetic resolution of (R/S )-­1-­phenylethanol.

2) Is there a biocatalytic route available? Once we identify a publication or set of publications for our desired transformation, the workflow is very similar to a classical organic synthesis approach. What reagents do we need? What type of equipment/instruments do we need? 3) Is there a commercial enzyme available? This is perhaps the less intuitive step for a classical synthetic chemist because often when browsing for commercial enzymes, one will find multiple formulations of the same enzyme (see example below). In this case study, the authors screened three different enzymes: Porcine pancreas lipase (PPL), lipase from Thermomyces lanuginous (TL-­IM), and lipase-­B from Candida antarctica (CAL-­B). Looking into Sigma Aldrich for Candida antartica lipase B, one will find multiple entries. We can simplify the search by filtering the results to only “enzymes” and now only four different CAL-­B formulations will appear (see Figure 1.24). In order to decide which of the multiple enzymes are the most appropriate for the desired transformation, one needs to evaluate several parameters, for instance, is enzyme re-­use desired? If yes, then the enzyme will need to be immobilized on a support (entries 2 and 3 from Figure 1.24). If not, one could go for the lyophilized powder (entry 1 from Figure 1.24). Analyzing the procedure from the original work of the case study, one can easily deduce that the enzymes used are immobilized since after the defined reaction time, and the enzymes are subsequently filtered. The options are now narrowed down to two; in this case the only difference between both formulations is the recombinant micro-­organism that was used as host for the enzyme synthesis (i.e. both enzymes are identical). If nothing is stated in the procedure (as in our case study), the simplest solution is to go for the formulation that contains the most units per enzyme (see explanation below). Thus, in our case and considering all the parameters analyzed above, the enzyme of choice would be “Lipase B Candida antarctica immobilized on Immobead 150, recombinant from yeast.”

1.7  Case Study: Setting up a Biotransformation

Figure 1.24  Sigma Aldrich search results for “Candida Antartica lipase B” (accessed 11 April 2023).

The same reasoning must be applied for the rest of the lipases. In any case, we should expect that, as with any published synthetic organic procedure, the commercially available reagents are reported in the materials and methods section or in the supplementary information (this case study).

29

30

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

To conclude, it is important to highlight that there is no single golden rule when deciding which biocatalyst/formulation/supplier that will a priori work best. Therefore, sometimes a broader screening is necessary to find the best candidate for a specific reaction. This task does not differ from the screening of, for example, a Pd/C catalyst within different suppliers with different loadings and activations in organic synthesis. 4) Choosing an enzyme based on your needs In our published case study, the rationale for their choice of enzyme is due to the fact that they are known to be more stable when exposed to organic solvents [87]. Additionally, having the enzymes immobilized enables future re-­ usage of the catalyst. 5) Perform a standard activity assay Let us take the example of “Lipase B Candida antarctica immobilized on Immobead 150, recombinant from yeast.” The product description contains a section where it states the unit definition. In this case, 1 U corresponds to the amount of enzyme (mg) needed to produce 1 μmol of butyric acid per minute at pH 7.5 and 40 °C, using glyceryl tributyrate as starting material. In this case, the supplier does not have a step-­by-­step protocol for the activity assay. However, one can find several suitable activity assay protocols for lipase to perform, depending on the equipment available in the lab, in a review published by Hasan et al. [88]. One can readily notice that this reaction is opposite to the one that we want to carry out (acylation). For an organic chemist, this may be counterintuitive since the natural way of testing the activity of a catalyst is in the reaction direction that one wants to perform. This does not mean that in the biocatalytic synthesis, the “real” reaction is not tested, but that a standard reaction needs to be performed prior to the desired transformation, to ensure that the biocatalyst has not lost activity over time due to, for instance, exposure to temperature, organic solvents, or high salt concentrations. Generally, before first use of the enzyme for experimentation, a standard activity assay using the conditions described by the supplier should be performed first to determine the baseline activity. This ensures that in the future, one can easily detect if the enzyme has lost activity over time and, second, to ensure that the purchased enzyme is complying with the specifications provided by the supplier. 6) Test your own substrates This part is probably the most important and interesting section of the whole process. Here, standard organic chemistry techniques are used. One simply needs to design reaction conditions based on a reported procedure and adjusting accordingly. Reactions are run exactly in the same manner as a synthesis

  ­Reference

procedure (weighing starting materials, calculating stoichiometries, monitoring reaction progress using chromatographies). Once initial results are obtained, one can then screen various conditions to achieve optimal parameters for high yields/conversions, enantioselectivity, reaction time, etc.

1.8  ­Concluding Remarks Over the past few sections, we gave insight into the feasibility of introducing biocatalysis in a standard chemistry lab. We hope we have instilled the reader with enough foundational knowledge to take their first steps into establishing biocatalytic reactions in their laboratory. Taking that first step toward discovering the power of enzymes for chemical synthesis will allow the reader to not only see the ease of incorporating enzymes into the workflow but also give one ­confidence in trying to expand the toolbox of available enzymes for alternative transformations. Furthermore, we provide several references below for more advanced reading.

Additional Resources ●● ●●

●●

●● ●●

Enzymatic parameters and reporting results [34, 89] Advanced biocatalysis book with an overview of typical reactions and industrial biotransformations [90] “The Hitchhiker’s guide to biocatalysis: recent advances in the use of enzymes in organic synthesis” [91] Biotransformations in Organic Chemistry [92] The Transformative Power of Biocatalysis in Convergent Synthesis [93]

­References 1 Alcántara, A.R., Domínguez de María, P., Littlechild, J.A. et al. (2022). ChemSusChem 15: e202102709. 2 Winkler, C.K., Schrittwieser, J.H., and Kroutil, W. (2021). ACS Cent. Sci. 7 (1): 55–71. 3 Cox, R.M., Wolf, J.D., and Plemper, R.K. (2021). Nat. Microbiol. 6 (1): 11–18. 4 Sheahan, T.P., Sims, A.C., Zhou, S. et al. (2020). Sci. Transl. Med. 12 (541): eabb5883. 5 McIntosh, J.A., Benkovics, T., Silverman, S.M. et al. (2021). ACS Cent. Sci. 7 (12): 1980–1985.

31

32

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

  6 Corey, E.J., Bakshi, R.K., and Shibata, S. (1987). J. Am. Chem. Soc. 109 (18): 5551–5553.   7 Richard, J.P. (2013). Biochemistry 52 (12): 2009–2011.   8 Barrios-­Rivera, J., Xu, Y., Wills, M., and Vyas, V.K. (2020). Org. Chem. Front. 7 (20): 3312–3342.   9 Mitsukura, K., Suzuki, M., Tada, K. et al. (2010). Org. Biomol. Chem. 8 (20): 4533–4535. 10 Roura Padrosa, D., Benítez-­Mateos, A.I., Calvey, L., and Paradisi, F. (2020). Green Chem. 22 (16): 5310–5316. 11 Matzel, P., Wenske, S., Merdivan, S. et al. (2019). ChemCatChem 11 (17): 4281–4285. 12 Schmid, A., Dordick, J.S., Hauer, B. et al. (2001). Nature 409 (6817): 258–268. 13 Wu, S., Snajdrova, R., Moore, J.C. et al. (2021). Angew. Chem Int. Ed. 60 (1): 88–119. 14 Stewart, J.D. (2001). Curr. Opin. Chem. Biol. 5 (2): 120–129. 15 Contente, M.L., Roura Padrosa, D., Molinari, F., and Paradisi, F. (2020). Nat. Catal. 3 (12): 1020–1026. 16 Xu, J., Peng, Y., Wang, Z. et al. (2019). Angew. Chem Int. Ed. 58 (41): 14499–14503. 17 Crawshaw, R., Crossley, A.E., Johannissen, L. et al. (2021). Nat. Chem. 14 (3): 313–320. 18 Hollmann, F., Opperman, D.J., and Paul, C.E. (2021). Angew. Chem Int. Ed. 60 (11): 5644–5665. 19 Lee, M.-­Y. and Dordick, J.S. (2002). Curr. Opin. Biotechnol. 13 (4): 376–384. 20 Ni, Y., Holtmann, D., and Hollmann, F. (2014). ChemCatChem 6 (4): 930–943. 21 Vieille, C. and Zeikus, G.J. (2001). Microbiol. Mol. Biol. Rev. 65 (1): 1–43. 22 Carrea, G. and Riva, S. (2000). Angew. Chem Int. Ed. 39 (13): 2226–2254. 23 Klibanov, A.M. (2001). Nature 409 (6817): 241–246. 24 Zaks, A. and Klibanov, A.M. (1986). J. Am. Chem. Soc. 108 (10): 2767–2768. 25 Arnold, F.H. (1990). Trends Biotechnol. 8 (C): 244–249. 26 Cui, H., Vedder, M., Schwaneberg, U., and Davari, M.D. (2002). Using Molecular Simulation to Guide Protein Engineering for Biocatalysis in Organic Solvents. In: Enzyme Engineering. Methods in Molecular Biology, vol. 2397 (ed. F. Magnani, C. Marabelli, and F. Paradisi). New York, NY: Humana. 27 Di Cera, E. (2006). J. Biol. Chem. 281 (3): 1305–1308. 28 Schmidt, J., Wei, R., Oeser, T. et al. (2016). FEBS Open Bio 6 (9): 919–927. 29 Albe, K.R., Butler, M.H., and Wright, B.E. (1990). J. Theor. Biol. 143 (2): 163–195. 30 Turgay Yagmur, I., Unal Uzun, O., Kucukcongar Yavas, A. et al. (2020). Ann. Allergy Asthma Immunol. 125: 460–467. 31 Dörr, T., Moynihan, P.J., and Mayer, C. (2019). Front. Microbiol. 10: 2051. 32 Amarasiri, M., Sano, D., and Suzuki, S. (2020). Crit. Rev. Environ. Sci. Technol. 50 (19): 2016–2059.

  ­Reference

33 Katouli, M. (2010). Iran. J. Microbiol. 2 (2): 59–72. 34 Bisswanger, H. (2014). Perspect. Sci. 1 (1–6): 41–55. 35 Labuda, J., Bowater, R.P., Fojta, M. et al. (2018). Pure Appl. Chem. 90 (7): 1121–1198. 36 Units of Enzyme Activity (1979). Eur. J. Biochem. 97 (2): 319–320. 37 Sofer, W. and Martin, P.F. (1987). Annu. Rev. Genet. 21 (1): 203–225. 38 Sellés Vidal, L., Kelly, C.L., Mordaka, P.M., and Heap, J.T. (2018). Biochim, Biophys. Acta – Proteins Proteomics 1866 (2): 327–347. 39 Rej, R. and Shaw, L.M. (1984). CRC Crit. Rev. Clin. Lab. Sci. 21 (2): 99–186. 40 Solares, L.F., Brieva, R., Quirós, M. et al. (2004). Tetrahedron Asymmetry 15 (2): 341–345. 41 Camm, E.L. and Towers, G.H.N. (1973). Phytochemistry 12 (5): 961–973. 42 Read, J., Pearce, J., Li, X. et al. (2001). J. Mol. Biol. 309 (2): 447–463. 43 Utter, M.F. and Keech, D.B.J. (1960). Biol. Chem. 235 (5): PC17–PC18. 44 Holliday, G.L., Rahman, S.A., Furnham, N., and Thornton, J.M. (2014). J. Mol. Biol. 426 (10): 2098–2111. 45 Alegria-­Schaffer, A. (2014). Methods Enzymol. 541: 251–259. 46 Wei, H., Wang, Z., Zhang, J. et al. (2011). Nat. Nanotechnol. 6 (2): 93–97. 47 Sanghamitra, N.J.M. and Ueno, T. (2013). Chem. Commun. 49 (39): 4114–4126. 48 Fang, P., Liu, M., Xue, Y. et al. (2015). Analyst 140 (22): 7613–7621. 49 Lund, I.T., Bøckmann, P.L., and Jacobsen, E.E. (2016). Tetrahedron 72 (46): 7288–7292. 50 Mouad, A.M., Taupin, D., Lehr, L. et al. (2016). Mol. Catal. B Enzyme 126: 64–68. 51 Hasegawa, S., Azuma, M., and Takahashi, K. (2008). J. Chem. Technol. Biotechnol. 83 (11): 1503–1510. 52 Mei, Y., Kumar, A., and Gross, R.A. (2002). Macromolecules 35 (14): 5444–5448. 53 Mateo, C., Palomo, J.M., Fernandez-­Lorente, G. et al. (2007). Enzyme Microb. Technol. 40 (6): 1451–1463. 54 Idris, A. and Bukhari, A. (2012). Biotechnol. Adv. 30 (3): 550–563. 55 Ortiz, C., Ferreira, M.L., Barbosa, O. et al. (2019). Catal. Sci. Technol. 9 (10): 2380–2420. 56 Hvidsten, I.B. and Marchetti, J.M. (2021). Energy Convers. Manag. X 10: 100061. 57 Jones, H. and Venables, W.A. (1983). FEBS Lett. 151 (2): 189–192. 58 Akita, H., Suzuki, H., Doi, K., and Ohshima, T. (2014). Appl. Microbiol. Biotechnol. 98 (3): 1135–1143. 59 Bhat, M.K. (2000). Biotechnol. Adv. 18 (5): 355–383. 60 Senior, A.W., Evans, R., Jumper, J. et al. (2020). Nature 577 (7792): 706–710. 61 Finnigan, W., Hepworth, L.J., Flitsch, S.L., and Turner, N.J. (2021). Nat. Catal. 4 (2): 98–104. 62 Mándity, I.M., Ötvös, S.B., and Fülöp, F. (2015). ChemistryOpen 4 (3): 212–223. 63 Yoshida, J.I., Kim, H., and Nagaki, A. (2011). ChemSusChem 4 (3): 331–340.

33

34

1  Biocatalysis 101 – A Chemist’s Guide to Starting Biocatalysis

64 Rahman, M.T. and Wirth, T. (2018). Safe use of hazardous chemicals in flow. In: Flow Chemistry for the Synthesis of Heterocycles. Topics in Heterocyclic Chemistry, vol. 56 (ed. U. Sharma and E. Van der Eycken). Cham: Springer. 65 Benítez-­Mateos, A.I., Contente, M.L., Roura Padrosa, D., and Paradisi, F. (2021). React. Chem. Eng. 6 (4): 599–611. 66 Britton, J., Majumdar, S., and Weiss, G.A. (2018). Chem. Soc. Rev. 47 (15): 5891–5918. 67 Santi, M., Sancineto, L., Nascimento, V. et al. (2021). Int. J. Mol. Sci. 22 (3): 1–32. 68 Khan, M.R. (2021). Bull. Natl. Res. Cent. 45: 1. 69 Thangaraj, B. and Solomon, P.R. (2019). ChemBioEng Rev. 6 (5): 157–166. 70 Sheldon, R.A., Basso, A., and Brady, D. (2021). Chem. Soc. Rev. 50 (10): 5850–5862. 71 Sheldon, R.A. and Pelt, S.V. (2013). Chem. Soc. Rev 42: 6223. 72 Rodrigues, R.C., Berenguer-­Murcia, Á., Carballares, D. et al. (2021). Biotchnol. Adv. 52: 107821–107858. 73 Romero-­Fernández, M. and Paradisi, F. (2020). Curr. Opin. Chem. Biol. 55: 1–8. 74 Wenger, L., Radtke, C.P., Göpper, J. et al. (2020). Front. Bioeng. Biotechnol. 8: 713. 75 Schmieg, B., Döbber, J., Kirschhöfer, F. et al. (2019). Front. Bioeng. Biotechnol. 6 (JAN): 211. 76 Maier, M., Radtke, C.P., Hubbuch, J. et al. (2018). Angew. Chem Int. Ed. 57 (19): 5539–5543. 77 Mazurenko, S., Prokop, Z., and Damborsky, J. (2020). ACS Catal. 10 (2): 1210–1223. 78 Yi, D., Bayer, T., Badenhorst, C.P.S. et al. (2021). Chem. Soc. Rev. 50 (14): 8003–8049. 79 Huang, P.S., Boyken, S.E., and Baker, D. (2016). Nature 537 (7620): 320–327. 80 Xiong, W., Liu, B., Shen, Y. et al. (2021). Biochem. Eng. J. 174: 108096. 81 Schmermund, L., Jurkaš, V., ss et al. (2019). ACS Catal. 9 (5): 4115–4144. 82 Hyster, T.K. (2020). Synlett 31 (03): 248–254. 83 Schmermund, L., Jurkaš, V., Özgen, F.F. et al. (2019). ACS Catal. 9 (5): 4115–4144. 84 Chanquia, S.N., Valotta, A., Gruber-­Woelfler, H., and Kara, S. (2021). Front. Catal. 1: 816538. 85 Lee, S.H., Choi, D.S., Kuk, S.K., and Park, C.B. (2018). Angew. Chem Int. Ed. 57 (27): 7958–7985. 86 Bandeira, P.T., Thomas, J.C., De Oliveira, A.R.M., and Piovan, L. (2017). J. Chem. Educ. 94 (6): 800–805. 87 Stepankova, V., Bidmanova, S., Koudelakova, T. et al. (2013). ACS Catal. 3 (12): 2823–2836. 88 Hasan, F., Shah, A.A., and Hameed, A. (2009). Biotechnol. Adv. 27 (6): 782–798. 89 Halling, P.J. and Gupta, M.N. (2014). Perspect. Sci. 1 (1–6): 98–109.

  ­Reference

90 Tibhe, J.D. (2012). Green Process. Synth. 1: 6. 91 Sheldon, R.A., Brady, D., and Bode, M.L. (2020). Chem. Sci. 11 (10): 2587–2605. 92 Faber, K. (2011). Biotransformations in Organic Chemistry, 6ee, 315–324. Heidelberg: Springer. 93 Zetzsche, L.E., Chakrabarty, S., and Narayan, A.R.H. (2022). Am. Chem. Soc. 144: 5214–5225.

35

37

2 Introduction to Photochemistry for the Synthetic Chemist Stefano Protti, Davide Ravelli, and Maurizio Fagnoni PhotoGreen Laboratory, Department of Chemistry, University of Pavia, Pavia, Italy

Glossary Absorbance (A)  Absorbance is defined (and calculated) as the logarithm of the ratio of incident to transmitted radiant power (light) through a sample. Chromophore  The part (atom or group of atoms) of a molecular entity in which the electronic transition responsible for a given spectral band is approximately localized. The term arose in the dyestuff industry, referring originally to the groupings in the molecule that are responsible for the dye’s color. Excited state  State of a system with energy higher than that of the ground state. This term is most commonly used to characterize a molecule in one of its electronically excited states but can also refer to vibrational and/or rotational excitation in the electronic ground state. Ground state  The state of lowest Gibbs energy of a system. HOMO  Highest-­energy filled (or partially filled) molecular orbital of a molecular entity. Intersystem Crossing (ISC)  A photophysical process. An isoenergetic radiationless transition between two electronic states having different multiplicities. It often results in a vibrationally excited molecular entity in the lower electronic state, which then usually deactivates to its lowest vibrational level. For further information on photochemical terminology, please refer to https://goldbook. iupac.org/. Enabling Tools and Techniques for Organic Synthesis: A Practical Guide to Experimentation, Automation, and Computation, First Edition. Edited by Stephen G. Newman. © 2023 John Wiley & Sons, Inc. Published 2023 by John Wiley & Sons, Inc.

38

2  Introduction to Photochemistry for the Synthetic Chemist

LUMO  Lowest-­energy unoccupied (completely or partially vacant) molecular orbital of a molecular entity. (Molar) Extinction coefficient (ε)  The linear absorption coefficient divided by the amount concentration. Photocatalysis  A change in the rate of a chemical reaction, or its initiation due to the action of a radiation in the presence of a chemical entity (dubbed as photocatalyst) that absorbs light and is involved in the chemical transformation of the reaction partners. Photochemical reaction  Generally used to describe a chemical reaction caused by absorption of ultraviolet, visible, or infrared radiation. There are many ground-­state reactions, which have photochemical counterparts. Among these are photochemical nitrogen extrusions, photocycloadditions, photodecarbonylations, photodecarboxylations, photoenolizations, photo-­ Fries rearrangement, photoisomerizations, photooxidations, photorearrangements, photoreductions, photosubstitutions, etc. Photon  Particle of zero charge, zero rest mass, spin quantum number 1, energy hν and momentum hν/c (h is the Planck constant, ν the frequency of radiation and c the speed of light), carrier of electromagnetic force. Photosensitized reaction  A photochemical or photophysical alteration that takes place in one molecular entity as a result of light absorption by another molecular entity defined as “photosensitizer.” Singlet state  A spin state that shows a total electron spin quantum number equal to 0. Transmittance (T)  The ratio of the transmitted radiant power (P) to that incident on the sample (P0). T = P/P0. Triplet state  A spin state that shows a total electron spin quantum number of 1.

2.1 ­Introduction 2.1.1  Light to Make Your Synthesis Greener In common organic chemistry textbooks, most of the reactions considered ­useful for practical applications make use of “thermal” conditions, namely taking place through processes that simply require mixing the reagents in the presence of an additive (e.g. an acid, a base, or an (organo)catalyst) at a suitable temperature (usually involving [mild] heating). However, the concerns raised by the pollution connected with chemical processes have recently pushed chemists to (re)consider alternative approaches, until now limited primarily to academic studies. Thus, different strategies have been proposed to surmount this problem, and

2.1 ­Introductio

photochemistry, which studies the use of light to promote a chemical ­conversion, is widely ­considered as one of the best choices. In fact, it is reasonable to speculate that green chemistry and photochemistry were born at the same time [1]. A pioneer in this field was Giacomo Ciamician and his work at the beginning of the twentieth century deeply contributed to the foundation of both photochemistry and sustainable chemistry  [2]. He recognized that an organic reaction may be carried out under conditions milder than those typical of thermal processes by using a ­renewable source of energy, such as solar light, to mimic what plants realize in everyday life, namely converting luminous light to chemical bonds. In fact, he stressed that “using aggressive reagents and high temperatures is almost always unavoidable when carrying out an organic synthesis in the laboratory. Deploying energy would, on the other hand, not be so frustrating for modern organic chemistry, were it not that the living world, in particular plants, gives us the marvelous example of great results obtained, at least from what appears, by using minimal means” [2]. During the last decades, it has become clear that the ideal traceless reagent able to promote a chemical reaction leaving no residues behind is the photon  [3, 4]. The concept that photochemistry uses the energy of  the photon only as a simple heating surrogate, however, is a trivial simplification.

2.1.2  A Way to Overcome HOMO/LUMO Interactions It is commonsense that the nature and the energy distribution of the molecular orbitals (in particular for what concerns frontier orbitals, namely HOMO and LUMO) dictates the chemical behavior of a given compound under thermal conditions. Thus, the electrophilicity of the carbonyl group or of an alkyl halide is due to the π* (C═O) or the σ* (C─X) orbitals, respectively, whereas the nucleophilicity of an alkene is set by the π (C═C) orbital. The outcome of a photochemical reaction, on the contrary, depends on the somewhat different electronic distribution of the reacting species (the “excited state”) with respect to the common chemistry stemming from the “ground state,” as detailed in the following. The fundamental law of photochemistry (the so-­called Grotthuss–Draper law) says that light must be absorbed by a chemical substance for a photochemical reaction to take place. The absorption of light by a chemical entity is due to the incorporation of a “chromophore” (e.g. a carbonyl group, a double bond, an aromatic ring, or an azo group). In order to have a chemical modification, the organic compound must absorb light, generally in the UV (wavelength of 100–400 nm) or/and visible (400–800 nm) range. The UV spectrum may be roughly divided into three regions, namely UV-­A (320–400 nm), UV-­B (280–320 nm), and UV-­C (100–280 nm). However, for a practical use the wavelength should be no shorter than 254 nm due to the lack of convenient light sources (vide infra).

39

40

2  Introduction to Photochemistry for the Synthetic Chemist

The absorption of UV and visible light by an organic molecule causes the promotion of an electron from an occupied molecular orbital (σ, π, or a nonbonding orbital, n) to an unoccupied molecular orbital (σ*, π*) forming an “electronically excited state.” The most common case is the excitation of an electron from the HOMO to the LUMO orbital leading to a species having two unpaired electrons in two distinct orbitals, mimicking a “diradical species” where a bond is weakened by the presence of an electron in the antibonding orbital. The two electrons in the excited state may accommodate in two ways, viz. with the spin anti-­parallel or parallel to each other, thus forming the corresponding singlet or triplet states, respectively. Singlets are directly accessed by excitation, whereas triplets may be obtained by Intersystem Crossing (ISC) from the singlets or via energy transfer with a triplet excited state of a suitable photosensitizer (see the following sections for further details). Usually both spin states lead to the same product, though exceptions exist, and a spin-­selective chemistry may take place [5]. The excited states are short-­lived species (in the μs range for triplets and ns for singlets) and may undergo a rapid physical decay, back to the more stable ground state [6]. Accordingly, a photochemical reaction must be by necessity faster than decay to avoid electronic deactivation. An organic compound, when in the excited state, shows a chemical reactivity markedly different from that of ground-­state molecules, taking advantage of the extra amount of energy provided by the photon absorbed. This contrasts with thermal reactions that may encounter high barriers to surmount. This was clearly stated by Prof. Noiory who highlighted how photochemistry “enhances the power of chemical synthesis by removing current thermodynamic restrictions” [7], thus representing an appealing tool for organic synthesis. Figure  2.1 shows some paradigmatic cases of excited states. Alkanes may be involved in a σ→σ* transition to give a σσ* state, but this is poorly useful since this transition requires high-­energy photons. The n→σ* transition in alkyl halides may have some applications (e.g. in the cleavage of C─X bonds); however, the π→π* transition in alkenes (or polyenes) and aromatics along with the π→π* (or n→π*) transition in carbonyl derivatives account for most photochemical transformations of synthetic interest. For a given compound, the excited state is an “electronic isomer” of the ground state, and this completely changes the known behavior of the relevant functional group typically studied in organic chemistry courses. The other side of the medal is that photochemistry may appear somewhat “magic” and, as stated by George Büchi, “the course of the transformations could rarely be predicted, thus robbing the investigator of the pleasure derived from designing new reactions” [8]. In fact, the introduction of different substituents in a given compound can interfere with the nature and the lifetime of the involved excited states, thus affecting the reaction course. Nonetheless, these changes can be studied and rationalized by detailed photophysical investigations.

2.1 ­Introductio

Molar extinction coefficient (M–1cm–1)

UV-Vis spectrum of Reagent εMAX

Reagent Ground state hv Reagent * Excited state

λMAX Product

Wavelength (nm) Chromophore

Transition

λMAX (nm)

εMAX (M–1cm–1)

CH3–CH3

σ→σ*

133

10 000

Double C=C bond

π→π*

165

1600

CH3–Cl

n→σ*

173

200

CH3–OH

n→σ*

177

200

Triple C≡C bond

π→π*

185

2000

CH3–I

n→σ*

259

400

Acetone

n→π*

280

12

Figure 2.1  Transitions of selected chromophores.

How light is absorbed is crucial. The Lambert–Beer law states that the ­absorbance of a solution is A = εbc, where c is the molar concentration (M) and b the optical path (cm). The value of εMAX (M−1 cm−1) (see Figure 2.1 above) is an indication of the feasibility of a transition, being low for a “partially forbidden” band (e.g. ca. 10–100 M−1 cm−1 for the nπ* band of aliphatic ketones at 280 nm) and high (at least 104 M−1 cm−1) for an “allowed” transition (e.g. the ππ* band in polyenes, ­aromatics, and dyes). Table 2.1 collects the longest wavelength where selected reagents have an absorbance A = 1 at a 0.01 M concentration, as a rough guide in the choice of a suitable λ for excitation. The two most important functional groups that are commonly involved in direct photochemical processes are the C═C bond in alkenes (dienes) and the C═O bond in ketones. Schemes 2.1a–e describe some of the most important photochemical reactions arising from these two chromophores. The knowledge of the peculiar behavior of the functional group when in the excited state is mandatory to design a selective photochemical reaction. The reaction occurring depends on the promoted electronic transition and, accordingly, on the excited state populated. However, in alkenes the promotion of an electron from a bonding orbital (π) to the

41

42

2  Introduction to Photochemistry for the Synthetic Chemist

Table 2.1  Absorption of selected organic compounds. Reagent

λa (nm)

Reagent

λa (nm)

Acetanilide

288

2,5-­Dimethyl-­2,4-­hexadiene

275

Acetophenone

304

Fluorobenzene

270

Acridine

395

Indene

291

Allyl alcohol

205

Iodobenzene

280

Allyl amine

232

Nitrobenzene

350

Allyl bromide

253

Phenanthrene

345

Aniline

308

Phenol

285

Anthracene

378

Pyrrole

238

Benzaldehyde

298

Quinine

370

Benzophenone

360

Retinol (Vitamin A)

395

Bromobenzene

276

Stilbene

333

Carbazole

350

1,1,2,2-­Tetrachloroethylene

260

Chlorobenzene

275

Thiophenol

280

2-­Cyclohexenone

310

Uracil

285

cis-­1,2-­Dichloroethylene

230

a

 The longest wavelength at which the reagent has absorbance A = 1 at a 0.01 M concentration (optical path 1 cm).

corresponding antibonding orbital (π*) generates a ππ* state, where formally the double bond is not present anymore and the rigidity of the C═C bond is removed (Figure 2.2, left part). As a result, an easy E to Z (or vice versa) conversion may take place. The reaction is feasible even when using azoderivatives, having a N═N rather than a C═C bond (Scheme 2.1a). The diradical nature of the excited C═C bond (especially in enones) is likewise useful for the construction of four membered rings by reaction with other alkenes via a [2+2] cycloaddition process (Scheme 2.1b). Likewise, the chemistry of the carbonyl group is largely affected by excitation (Figure  2.2, right part). In fact, the ground-­state chemistry is dominated by the “basicity” of the n orbitals (functioning as the HOMO) and the electrophilicity imparted by the π* orbital (the LUMO). Considering the n→π* transition, the resulting excited state (being diradical in nature) mimics, however, the behavior of alkoxyl radicals [9]. These radicals are known to fragment to restore the carbonyl group [10] or to abstract hydrogen atoms to form a strong O─H bond [11]. This is reflected in the C(═O)─Cα bond fragmentation to form a radical couple (Norrish type I reaction), from which CO may be lost to form a photoextrusion product

2.1 ­Introductio

(a)

R2

hv

A A

(b)

*

R2

R1

A A R1 ππ*

R1

hv′

R2

A A

O

O

X

hv

+

X

A = CH, N

(c) R1

O C

O C 2 1 R R nπ*

hv R2

*

O C + R2 R1

– CO

(e)

(d) O H R α

γ

hv

O

R

OH

HO

R

R1

O

+ R2

R1 R2

β

(f) O O

R1–R2

(g) 1O 2

1O 2

OOH

1O 2

O O

Scheme 2.1  Selected photochemical reactions of double bonds and carbonyls: (a) E/Z isomerization, (b) [2+2] photocycloaddition, (c) Norrish type I reaction, (d) Norrish–Yang reaction, (e) Paternò–Büchi reaction; and those involving singlet oxygen: (f) Schenk-­ene reaction (right) and synthesis of 1,2-­dioxetanes (left); (g) synthesis of 3,6-­dihydro-­1,2-­dioxines via a hetero-­Diels-­Alder reaction. Figure 2.2  Electronic transitions occurring in alkenes (π→π*) and ketones (n→π*).

Ketones

Alkenes π∗

π∗

π∗

π∗

hv

hv

n π

π C C

C C ππ∗ state

π

n π

C O

C O nπ∗ state

(Scheme  2.1c). Alternatively, an intramolecular hydrogen abstraction from the γ-­position may lead to a biradical intermediate, which upon intramolecular radical coupling forms a cyclobutanol (the Norrish–Yang reaction, Scheme 2.1d). In analogy with the C═C bond, the C═O group could be engaged in a [2+2] cycloaddition reaction to form oxetanes (the Paternò–Büchi reaction, Scheme 2.1e) [12]. In the last decade, photochemistry strongly emerged also as a tool for the  ­photogeneration of reactive ground-­state intermediates (carbon-­ and heteroatom-­based radicals, radical ions, diradicals, ions, carbenes, and

43

44

2  Introduction to Photochemistry for the Synthetic Chemist

nitrenes) [13–15]. Most intermediates can be easily detected and characterized by time-­resolved ­spectroscopic techniques (not treated in this chapter), such as Laser Flash Photolysis [16]. This is a precious tool in the understanding and engineering of a photochemical process, since, once the generation of a given intermediate has been confirmed under the employed reaction conditions, practitioners can then investigate how to govern the ensuing chemistry. These reactive intermediates may be photochemically formed in different ways. Typical examples are the photohomolysis [13–15, 17] or photoheterolysis [18] of chemical bonds, taking place with a negligible activation energy, provided that the energy of the photon is sufficient to achieve the excited state and promote the desired cleavage (Figure 2.3, path b). In most cases, only a given bond is broken, leading to a clean reaction. Interestingly, a reaction may derive from the excited state of a reactant that did not absorb light. The typical case is that of a photosensitized reaction, wherein a photosensitizer (PS) is responsible of light absorption and then its excited state (PS*) engages an energy transfer with the reactant, ultimately populating its excited state (Figure 2.3, path c) [19]. This is a common strategy for the generation of the triplet state of a reactant while avoiding the intermediacy (and the chemistry) of the singlet excited state that is usually formed upon direct irradiation of the same reactant. A particular Negligible Ea

Heat (∆), chemical activation

Intermediate

a

hv

b′

Products

Reactant

PS

Reactant*

b

Reactant

PS*

PCET* d

1O

3O 2

2

PS Oxidized products

hv

c

PCHAT* e

f

PS* Radicals or radical ions

Radicals

Figure 2.3  General scheme describing thermal (in magenta), photochemical (in gray), photosensitized (in red), and photocatalytic (in blue) processes occurring on a reactant. PC = Photocatalyst. PS = Photosensitizer. HAT = Hydrogen Atom Transfer. ET = Electron Transfer.

2.2  ­How to Plan a Photochemical Synthesi

case is the generation of the excited state of molecular oxygen. In fact, the ground state of oxygen is a triplet since it contains two unpaired electrons (3O2). Direct excitation of 3O2, however, is unpractical due to the prohibitive wavelengths required. Accordingly, the formation of singlet oxygen (1O2), a well-­known oxidizing agent, can be triggered through the adoption of a suitable photosensitizer (e.g. a dye; Figure 2.3, path d) [20–22]. Widely used is the reaction of 1O2 with alkenes to form 1,2-­dioxetanes or allylic hydroperoxides (the Schenk-­ene reaction, Scheme 2.1f) or with dienes to give 3,6-­dihydro-­1,2-­dioxines (Scheme 2.1g) [20–22]. In other instances, the reactive molecule neither absorbs light nor is its excited state involved in the reaction. This is the case of photocatalyzed reactions [23–30], wherein an excited photocatalyst (PCET* or PCHAT*) activates a light-­transparent reagent by a chemical reaction, usually involving an electron transfer (ET, Figure 2.3, path e) [23–30] or a hydrogen atom transfer (HAT, path f) step [31–33]. In the former case, the process takes advantage of the enhanced oxidation or reduction capability of PCET when in the excited state to generate radical ions (or radicals when charged substrates are used) upon reaction with an organic compound. In the latter case, PCHAT* mimics the alkoxy radical behavior and is thus able to cleave homolytically a C─H bond to form a carbon-­centered radical. Figure 2.4 collects representative photocatalysts/photosensitizers currently used in synthesis, some of them being simple organic compounds (PhotoOrganoCatalysts, POC) [34–37].

2.2 ­How to Plan a Photochemical Synthesis A crucial issue in the design of a photochemical reaction is assuring an efficient light absorption of the photoreacting species [38]. In this respect, the preliminary measurement of the UV-­visible spectrum of the reaction mixture components and the identification of the major absorption bands is mandatory. Later on, other factors must be taken into account as detailed in the following.

2.2.1  The Choice of the Solvent An important parameter when planning a photochemical process is the choice of the solvent. Apart from the obvious issue of the solubility of reagents/additives, the solvent must not absorb competitively with the photoreacting species. Figure 2.5 gives an indication on the transparency of common solvents, showing that solvents having different polarities can be conveniently chosen. Having a look at the absorption edge for these solvents, it is apparent that acetonitrile and alcohols (MeOH or EtOH) are ideal polar solvents, whereas alkanes (when deprived of other hydrocarbon impurities) and ethers may be suitable as apolar or medium polar media, respectively. All solvents reported are transparent in the visible region.

45

Cl

O

N

Ph2CO BP

Me2N

NMe

S

O AQ

MB

2

Cl

Cl

Cl

CO2H

I

I

+

HO

O

O

I

I RB

CO2H Br

Br HO

N

+ N Me

O

O Br

N

Br

Mes-Acr+

EY

N RuII N

N N N

N

RuII(bpy)32+

Ir

N

fac-IrIII(ppy)3

O O W W O O O O O O OO O O W O W O W O O W O O OW O W O O O O OO O O OW O W O O DT = [W10O32]4–

Figure 2.4  Common photocatalysts/photosensitizers used in synthetic applications. BP = benzophenone; AQ = anthraquinone; MB+ = methylene blue; RB = rose bengal; EY = eosin Y; Mes-­Acr+ = 9-­mesityl-­10-­methylacridinium salt; RuII(bpy)32+ = tris-­(2,2′-­bipyridine) ruthenium; fac-­IrIII(ppy)3 = fac-­(tris-­(2,2′-­phenylpyridine))iridium; DT = decatungstate anion. 100

Transmittance (%)

80

60 Acetonitrile n-Hexane Methanol Dichloromethane DMF Acetone

40

20

0 200

250

300

350

400

Wavelength (nm)

Figure 2.5  Absorption edge of some commonly used solvents (1 cm optical path). Source: Albini and Fagnoni [13]/with permission from John Wiley & Sons.

2.2  ­How to Plan a Photochemical Synthesi

Albeit counterintuitive, a photochemical reaction may take place even when the reagent is in the solid state (e.g. by irradiating crystals) leading even to enantioselective processes [39, 40]. An alternative is the photolysis of a suspension of microcrystals obtained by precipitating the reagent in a proper solvent while ­stirring  [41]. In some instances, photocatalyzed reaction may be carried out under  heterogeneous conditions where the PC is not soluble in the reaction mixture [13].

2.2.2  Concentration of the Absorbing Species Other crucial issues in a photochemical reaction are the concentration and the molar absorption coefficient (which represents a measure of how strongly a compound absorbs light at a given wavelength) of the photoreacting species, which reflect the absorbance of the system. Roughly, adopting a high concentration lengthens the required irradiation time, in turn making the process sluggish. As an example, irradiating a 0.1 M solution of the reagent (vessel having a 1 cm path) on a wavelength maximum with εMAX = 10 M−1 cm−1, an A = 1 results. From the Lambert–Beer law, A = log(1/T), where T = transmittance, a measure of the fraction of light that passes through a material (see Glossary). Accordingly, 90% of the light intensity is absorbed by the solution, but 10% is transmitted (and is thus not useful for the transformation). If εMAX = 104 M−1 cm−1, however, 99.9% of the light is absorbed, since A  =  103, and this means that 90% of the photons are absorbed in the first 0.1 mm layer of the solution, while the remainder of the solution cannot interact with light, remaining unreactive. A balance is then required since a too low concentration requires a huge amount of solvent (and of solution, accordingly) for the obtainment of the required product(s) on the desired scale, while a too high concentration may call for a long irradiation time rendering the process unpractical. An effective mixing of the solution, however, can alleviate (at least, in part) such a drawback. A general improvement in this respect is the use of flow conditions [42–45] or microreactors [46–48] (vide infra), where the limited optical path of the reacting solution allows for an efficient absorption of light. A further problem may arise from the competitive absorption of the ­thus-­formed photochemical product(s) that could completely hinder the photoreaction, when photostable (the so-­called inner filter effect), or yield further undesired secondary photoproducts, if photoreactive. In some instances, the products (or by-­products, such as polymers) may form films on the wall of the vessel, thus stopping the reaction because the remaining reagent in solution receives little, if any, light. As apparent from the latter consideration, there is not a general rule for the design of a given photochemical process and, in each case, a fine tuning of the reaction conditions is required when investigating a new reaction.

47

2  Introduction to Photochemistry for the Synthetic Chemist

2.2.3  The Reaction Vessel A photochemical reaction may be roughly carried out in two ways: by external or internal irradiation. In the first case, the solution to be irradiated is confined in a vessel (usually a tube or a vial) and the reaction is promoted by an external light source (see Section 2.2.4). This situation is particularly suited for small-­scale photoreactions, where small amounts of solution (ca. 1 mL) may be used. This setup is indicated in the optimization of the various reaction parameters (concentration of the reagents and/or additives, best wavelength, etc.), avoiding the need to waste (expensive) starting materials. Furthermore, the vessel may be purposely designed, allowing to tune the scale of the reaction. As apparent from Figure 2.6, with a light source emitting at 254 nm only rather expensive vessels made of fused quartz (that are transparent above 170 nm) can be used. The transparence limit of Pyrex glassware (or borosilicate glass) is around 300–310 nm; albeit less expensive, these vessels (typical examples include vials or round-­bottomed flasks commonly available in a chemistry laboratory) are practical only when irradiating with λ > 350 nm, therefore including the entire range of visible light.

2.2.4  Light Sources A judicious choice of the light source is fundamental for increasing the efficiency of the reaction, and a good matching between the absorption band of the photoreacting species and the wavelength emitted by the light source is required, also 100

80 Transmittance (%)

48

60

Fused quartz Borosilicate glass “Optical” glass Microscope slide

40

20

0 200

250

300

350

400

450

400

Wavelength (nm)

Figure 2.6  Transparency of materials used for the reaction vessel (1 mm thick). Source: Albini and Fagnoni [13]/with permission from John Wiley & Sons.

2.2  ­How to Plan a Photochemical Synthesi

verifying that nothing else interferes (e.g. the wall of the vessel or the cooling jacket of the lamp [vide infra]) with the photons in reaching the target molecule. A ­photochemical system should be engineered considering the dimension and the shape of the lamp along with the geometrical arrangement with respect to the reactor vessel. In this regard, performing a photochemical reaction at the industrial level is still challenging, despite a hard work has been devoted in the setup of photochemical reactors suitable to promote an efficient synthesis on a large scale (kg) [49, 50]. As it has been previously stressed, in some cases the reagent may not be the light absorbing component, as in the case of photosensitized or photocatalyzed reactions. In rare instances, the solvent itself may be the absorbing species, as in the case of acetone when adopted as photosensitizer (see Glossary). The type of light sources used depends on the range of wavelength required for triggering the desired reaction. It is possible to find focalized sources, where the beam is directed on the object to be irradiated, or non-­focalized sources, where light is emitted over the entire sphere described by the light source, thus decreasing the light power that hits the target. In the UV range, the most largely applied light sources are mercury vapor arcs, which can be divided according to the ­operating pressure. The main drawback of these lamps is the low efficiency in converting the electrical power into light that is dispersed over a range of wavelengths and toward all directions. Nowadays, there are several firms supplying photochemical apparatuses including the light source [51], the appropriate power supply, and in some cases the reaction flask with accessories, e.g. for gas inlet. The main characteristics of lamps used for photochemical synthesis are presented in the following. 2.2.4.1  Low-­Pressure Mercury Arcs

Low pressure (ca. 10–5 atm under operating conditions) Hg arcs (often dubbed as germicidal lamps) are long-­lived (>104 hours) light sources traditionally used in organic photochemistry. These are supplied as quartz tubes of various lengths, typically from 20 to 60 cm and 6–16 W power (Figure 2.7a), exhibiting an emission at 254 nm. However, when the lamps are internally coated with a phosphor (or a combination of phosphors) that absorbs the Hg radiation, the lamp emission can be shifted to longer wavelengths. These light sources include: ●●

●●

●●

lamps with emission centered in the UV-­B region, at 305–310 nm (a white phosphor is used); “black light” (also called “wood light”) lamps (see again Figure 2.7a), painted with BaSi2O5  :  Pb+ phosphor (emitting at 350–355 nm) or a SrB4O7F  :  Eu2+ phosphor (emitting at 368–370 nm); lamps with the emission centered in the visible range at various wavelengths, e.g. in the blue around 450 nm or over the whole spectrum (“compact fluorescent lamps,” CFL).

49

50

2  Introduction to Photochemistry for the Synthetic Chemist (a)

(b)

Figure 2.7  (a) Low-­pressure mercury arcs used for photochemical reactions. From top to bottom: 350–370 nm phosphor-­coated “wood light” lamp, 305–310 nm phosphor-­coated lamp, and 254 nm emitting lamp. (b) A typical multilamp apparatus fitted with low pressure mercury lamps.

A multilamp apparatus where 8–12 Hg lamps are circularly disposed is shown in Figure 2.7b; the test tubes containing the solution to be irradiated are placed inside. A rotating “merry-­go-­round” setup ensures a uniform illumination of the vessels and a cooling fan maintains the temperature 0.5 mm, flow rate > 1 mL min−1) or microflow reactors (having micro-­sized channels usually 90%) in the visible and UV light ranges. Moreover, these polymers exhibit high flexibility and resistance to relatively high pressures, and are inert to strong acids and bases, as well as do not suffer from swelling phenomena. Roughly, the assembling of a flow photoreactor may be designed in two ways. The reactor could be built around the lamp assuring an internal irradiation (see the case of the apparatus shown in Figure 2.10, left) [42], where the solution is flown in a transparent tubing wrapped around the lamp. By using this apparatus, considerable volumes (up to several liters) of solution may be irradiated in a limited period of time [42, 58]. External irradiation is well suited for “lab-­on-­a-­chip” microreactors, where the solution flows into microchannels while being irradiated by a lamp placed in front of the reactor (Figure 2.10, right). More recently, the design and preparation of 3D-­printed microreactors  [46–48] for light-­driven organic synthesis was proposed [59]. However, apart from lab-­made apparatuses, different continuous flow devices are commercially available [60, 61]. A peculiar case deals with “falling film” photoreactors, wherein a solution from a reservoir is sprayed at the top of the (cooled) lamp by means of a pump [62]. Alternatively, a thin layer of solution may be made to fall in the proximity of the light source or on the wall of the cooling jacket. Lamp Pump

UV transparent tubing

Radiated light Lamp Syringe pump Microchannels

Figure 2.10  Reactors using internal (left) and external (right) irradiation. Source: Knowles et al. [42]/Beilstein Institute for the Advancement of Chemical Sciences/CC BY-­2.0.

53

54

2  Introduction to Photochemistry for the Synthetic Chemist

2.2.6  Preparation of the Sample As hinted above (Section  2.1.2), the lifetime of excited states is rather short (roughly lifetime ca.1 μs for triplets and ca.1 ns for singlets) and a photochemical process must be very fast to favorably compete with physical decay. This has a positive implication in reducing the effect of impurities present in solution (thus the extensive purification of reagents and solvents is not generally required) but a large excess of the reaction partner may be mandatory in case of bimolecular reactions for intercepting the short-­lived excited state. A peculiar case is represented by molecular oxygen, which may quench excited states at diffusion-­controlled rate. The oxygen present in many air-­equilibrated organic solvents can be estimated in the range 2–3 × 10–3 M (5 × 10–4 M in water), thus, enough to quench >90% of long-­lived triplets. In some cases, oxygen is purposely added to the reaction mixture to ascertain if a triplet excited state is involved in the process or not. Moreover, O2 may smoothly react also with radical or radical ion species formed to deliver oxidation products, thus diverting the expected reaction outcome. At any rate, the amount of oxygen dissolved drops by ca. three orders of magnitude when flushing the solution with an inert gas (nitrogen or, better, argon) for some minutes to ensure “deoxygenated” conditions. In particular cases, where oxygen exerts a profound effect, freeze-­pump-­thaw cycles, where the solution is frozen (thanks to the use of liquid nitrogen) and the dissolved gas is then removed under vacuum, are compulsory. This precaution may be skipped, however, thus further simplifying the experimental procedure if it is proven that oxygen has no appreciable effect on the reaction outcome. A great advantage is the negligible dependence of many photochemical ­processes on temperature, thus the reactions are routinely carried out at room temperature, unless the nature and/or the lifetime of the thus formed products (or ground state intermediates) call for a specific temperature control (e.g. lowering or increasing).

2.2.7  Safety Equipment The use of intense light sources (including solar light) may raise some safety issues. In fact, UV-­B radiation is dangerous for the skin, causing erythema, and UV-­C is genotoxic. Likewise, the exposure to a strong UV-­A (up to 400 nm) or visible light radiation may cause damage. For this aim, irradiation apparatuses must be placed in a protected area. Direct eyes exposure to short-­wavelength light must be avoided during the execution of the experiment by wearing purposely designed protective glasses. However, accidental exposure may occur, which might cause a quite annoying conjuctivitis. When using high-­pressure mercury arcs, ozone may be formed locally in a high concentration. In this case, it is highly recommended to flush nitrogen in the close

2.3  ­Selected Applications of Photochemical/Photocatalyzed Reaction

vicinity of the arc (to avoid a buildup of ozone concentration) or carrying out the reaction under a fume hood. In addition, a proper disposal of the mercury-­ containing exhausted lamps is always mandatory.

2.3 ­Selected Applications of Photochemical/ Photocatalyzed Reactions In the following, selected examples are presented and discussed with no particular emphasis to the “mechanistic” detail of the reactions [63, 64]. This section provides several recent (mostly in the last 10 years) examples of diverse photochemical reactions involving the most common chromophores, including demonstration of a range of representative activation modes described in Figure 2.3 (the absorbing species in each scheme is highlighted in red). In addition to a general description of the chemical reaction, an abbreviated procedure is reported to give the reader insights into the technical aspects of experimental setup. Lastly, key information about the equipment used is given.

2.3.1  Reactions Involving the C═C Double Bond Rotation around a C═C double bond is energetically disfavored. Therefore, ­alkenes can exist as non-­interconverting E/Z isomers. These species can have very different shapes and properties, the case of vision being a notable example of the profound impact of this type of isomerism [65]. Very recently, photo-­induced E/Z isomerization of enamines has been exploited to achieve the deracemization of α-­branched aldehydes [66]. As shown in Scheme 2.2, this approach capitalized on a chiral primary amine organocatalyst and a readily available photosensitizer. Ir(ppy3) (0.5 mol%) tBu

O H

Ph

Et

HNTf2 (10 mol%) NH2 Et (S)-2 20 W LED (400 nm) PhCOOH (3 mol%) MeCN, 0 °C, 1 h, Ar

rac-1 (0.67 M)

R*

N

N

H

H

Ir(ppy3), hv

R* H

3

Ph

N

O H

Ph

(R)-1 73%, 94% ee H Ph

4

Scheme 2.2  Deracemization of α-­branched aldehydes.

55

56

2  Introduction to Photochemistry for the Synthetic Chemist

In the process, the racemic aldehyde (rac-­1) reacted with the aminocatalyst (S)-­2 to preferentially deliver the E-­enamine 3. 400 nm LED excitation of the Ir(ppy)3 photosensitizer followed by energy transfer to 3 triggered the conversion to the Z-­isomer 4 that underwent facially selective protonation leading preferentially to (R)-­1. Noteworthy, this protocol demonstrated a remarkable functional group tolerance, including ketones. However, the presence of an α-­aromatic substituent in the starting aldehyde is a prerequisite to obtain a good performance. Typical procedure for the synthesis of (R)-­1: To an oven-­dried 10 mL Schlenk tube equipped with a magnetic stir bar was added (S)-­2/HNTf2 (10 mol%), Ir(ppy)3 (0.5 mol%) and benzoic acid (3 mol%). Freshly prepared rac-­1 (0.2 mmol) dissolved in 0.3 mL MeCN was then added into the tube under Ar atmosphere. The mixture was degassed for three times using standard freeze–thaw method and stirred at 0 °C under irradiation of 20 W 400 nm LED for one hour. After the completion of deracemization, the solution was purified by flash column chromatography to give (R)-­1 (73%). The ee value of the isolated aldehyde was determined by HPLC analysis. Experimental setup: A 20 W 400 nm LED fixed on an aluminum block was glued at the bottom of a crystallizing dish by silicone gasket and grease (Figure 2.11). The temperature was controlled by the ethanol bath and coolant circulation pump. Vitamin D analogues endowed with biological activity are commonly prepared through a photosensitized protocol exploiting an E/Z isomerization of the C═C bond. As an example, calcipotriol (8), routinely used for the treatment of psoriasis can be obtained by deprotection of silyl ether 7 in turn formed from the conjugated triene derivative 5 upon irradiation with UV light in the presence of a catalytic amount of 9-­acetyl anthracene (6) as the photosensitizer (Scheme 2.3) [67]. The reaction has been performed both under batch and flow conditions. In the

Figure 2.11  (Left) Photoreactor. (Middle) 0.2 mmol scale reaction setup. (Right) 5 mmol scale reaction setup. Source: Huang et al. [66], Reproduced with permission from American Association for the Advancement of Science – AAAS.

2.3  ­Selected Applications of Photochemical/Photocatalyzed Reaction OH

H 5, (0.08 M) TBDSO

OTBDS

TBDS = tert-butyl dimethylsilyl

OH

OH

H

H

H 6 (5.7 mol%) hv (UV light) MTBE, 10 °C

H

OTBDS

TBDSO O

7

TBAF THF, 60 °C

HO

H 8

OH

6

Scheme 2.3  Photosensitized synthesis of Calcipotriol 8.

former case, the reaction was slower with a lower productivity (1.64 min g−1 ­product vs. 0.72 min g−1 in flow) and led to much less pure product (65% vs. 80% yield, under batch and flow conditions, respectively). Typical procedure for the synthesis of 7: An oxygen-­free solution of 5 in MTBE (1 g/20 mL) containing 450 mg of 6 was photo-­isomerized under batch and flow ­conditions. In both setups, the lamp providing UV-­light (medium pressure mercury lamp doped with iron from Heraeus: TQ718 Z4, 800 W power supply), surrounded by a quartz lamp-­housing comprising inner and outer jackets for water cooling, was placed in the center of a standard immersion well photoreactor having closable inlet and outlet openings. The lamp was temperature-­controlled by cooling the light housing with cooling water (10 °C). The irradiation layer thickness of the photoreactor was 9.7 mm. The reaction was kept under nitrogen at all times. In the flow version, the flow was controlled to be in the range of 3600–4800 mL min−1. Very recently, the enantioselective preparation of bridged cyclobutanes via a crossed intramolecular [2+2] cycloaddition has been reported (Scheme 2.4) [68]. The reaction occurred between the tethered C═C bond of the allyloxy moiety within 2-­(alkenyloxy)cyclohex-­2-­enones 9a-­c and took place in the presence of a chiral-­at-­Rh complex chelating Lewis acid (10, 2 mol%) upon irradiation with ­visible light (437 nm). Thus, 9a-­c bound to the catalyst to form a chiral complex and, upon excitation, delivered the desired cyclobutanes 11a-­c containing an otherwise elusive bridged framework with an excellent control over the absolute stereochemistry. Notably, the obtained tricyclic photocycloaddition products could be further elaborated to give different cyclobutane-­containing derivatives. Typical procedure for the synthesis of 11a-­c: An oven-­dried Schlenk tube, equipped with a magnetic stir bar, was charged with 10 (2 mol%), the chosen 9a-­c and 10 mL of dry DCE. The reaction mixture was deoxygenated by three cycles of

57

58

2  Introduction to Photochemistry for the Synthetic Chemist O

O

O

[Rh]

R1 R2

tBu

S N Rh N S

O

(10, 2 mol%)

10 W LED (437 nm) DCE, r.t., 14 h

9a-c (0.01 M)

[Rh]+ =

+PF – 6

Me

R1 R2

H

H

11a, R1 = R2 = H, 86%, 92% ee b, R1 = R2 = Me, 65%, 84% ee c, R1-R2 = -(CH2)3-, 59%, 80% ee

N N Me tBu

Scheme 2.4  Synthesis of bridged cyclobutanes.

“freeze-­pump-­thaw” and stirred at room temperature under blue LED irradiation (λ = 437 nm) to afford the corresponding cyclobutanes 11a-­c after purification by flash chromatography. Experimental setup: Photochemical experiments using a LED (λ = 437 nm) were carried out in a Schlenk tube (diameter  =  1.0 cm) with a polished quartz rod as an optical fiber, which was roughened by sandblasting at one end (Figure  2.12). The roughed end must be completely submerged in the solvent during the reaction in order to guarantee optimal and reproducible irradiation conditions.

2.3.2  Reactions Involving the C═O Double Bond

Figure 2.12  Photochemical apparatus employed for the synthesis of 11a-­c. Source: Rigotti et al. [68], Reproduced with permission from Royal Society of Chemistry.

The Paternò–Büchi (PB) reaction is one of the earliest known photochemical reactions having a synthetic potential and involves the direct excitation of an (aromatic) carbonyl to promote a [2+2] cycloaddition onto an olefin to form an  oxetane  [12]. Very recently, PB cycloadditions have been applied to

2.3  ­Selected Applications of Photochemical/Photocatalyzed Reaction

dearomatization reactions. The tetrahydrooxeto[2,3-­b]indole core in compounds 13a-­e, bearing up to three contiguous all-­substituted stereocenters, was efficiently built from indoles 12a-­e and aromatic ketones (e.g. benzophenone) upon visible light (405 or 456 nm) photolysis in toluene (Scheme 2.5) or acetone [69]. The procedure was smoothly optimized also under microfluidic flow conditions. Typical procedure for the synthesis of 13a-­e: The chosen indole (12a-­e, 0.1 mmol, 0.1 M) and benzophenone (1.0 equiv.) were dissolved in toluene or acetone (1 mL) and the reaction mixture was bubbled with N2 for one minute. The vial containing the so-­prepared solution was then degassed and irradiated for 16 hours in the photoreactor illustrated in Figure 2.13. The desired product was obtained via purification by chromatography (when needed). G

G O +

N Boc

Ph

hv (405 nm) Ph

N2, Toluene

1 equiv.

Ph Ph O

N Boc d.r. > 20:1

12a-e (0.1 M)

13a, G = H, > 98% b, G = 2-F, > 98%, c, G = 4-OMe, 78%, d, G = 3-Me, 72% e, G = 4-CF3, >98%

Scheme 2.5  Paternò–Büchi reaction for the construction of the tetrahydrooxeto [2,3-­b]indole core. Figure 2.13  Photochemical apparatus employed for the preparation of 13a-­e. Source: Rigotti et al. [68], Reproduced with permission from Royal Society of Chemistry (Creative Commons Licence).

Mirror

Fan

Light source

59

60

2  Introduction to Photochemistry for the Synthetic Chemist

Experimental setup: Figure  2.13 shows the general setup of a batch reaction under 405 nm irradiation. The reaction mixture was placed in the middle of the photoreactor (at about 1.5 cm distance from the light source). To maintain a stable reaction temperature, one fan was placed close to the vials (25 ± 2 °C) and the temperature was controlled by a thermometer.

2.3.3  Reactions Involving a Photoinduced Homolysis N-­Chloroamines have been considered as visible light reactive substrates to be used in atom-­transfer radical addition (ATRA) reactions [70]. Recently, a continuous flow reactor equipped with a separation membrane was used for the preparation of N-­chloroamines (e.g. 14). Subsequently, these species were directly used for the photochemical functionalization of olefins 15a-­e to afford β-­chloroamines 16a-­e in up to 91% yield and a throughput of more than 20 g h−1 (Scheme 2.6) [71]. All the reactions have been carried out by means of a commercial continuous flow reactor consisting of a compact glass fluidic module (155 × 125 × 8 mm size, 0.4 mm channel size, 2.77 mL internal volume), encased within a high-­capacity heat exchange channel (20 mL volume). Interestingly, the light source can be tuned according to the reaction conditions (365 nm for direct irradiation and 405 nm when 9-­fluorenone is present as photosensitizer). Typical procedure for the flow synthesis of 16a-­e: All solutions (containing bleach, piperidine, and 15a-­e) were degassed by sparging with an argon balloon. The reactor was turned on (LEDs, pump, and thermostats) and the system was given ~10 minutes to equilibrate. The solution reservoirs (loops of tubings) were charged with the amine and the bleach solutions and injected into the reactor. To account for any dilution effects at the edges of the injected sample, only the central fraction (~1 mL for optimization, 5.54 or 8.31 mL for isolation) was collected. Experimental setup: Flow reactions were conducted in a commercial continuous flow reactor: Corning Advanced-­Flow Lab Photo Reactor. hv (365 nm) G NaOCl NH Separation

G

15a-e, (0.5 M) N

Cl

(14, 1.5 equiv. 2 M in PhCH3)

H2SO4 (1.5 equiv.) MeCN

Cl N 16a,G = H, 91% b, G = 4-Br, 88% c,G = 4-tBu, 46% d, G = 4-OMe, 91% e, G = 3,4-OMe, 58%

Scheme 2.6  Atom-­transfer radical addition (ATRA) reaction between an N-­chloroamine and olefins.

2.3  ­Selected Applications of Photochemical/Photocatalyzed Reaction

Arylazo sulfones (ArN2SO2CH3, 17a-­e) are compounds bearing a dyedauxiliary group (a substituent responsible for both the color and the photoreactivity of a given starting substrate) and found wide application in the visible light and sunlight-­driven formation of both Ar-­C and Ar-­heteroatom bonds [72, 73]. The arylation of various (hetero)arenes (Figure 2.14, top) was successfully conducted in a lab-­made solar microcapillary reactor (the “sunflow” reactor, Figure 2.14, bottom) in three different geographical locations (Germany, Italy, and Brazil) with a short exposition time (one hour) [74]. Typical procedure for the flow synthesis of 18a-­e: The chosen arylazo sulfone (17a-­e, 0.2 mmol, 1.0 equiv., 0.05 M) was flushed with argon in a bottom flask and dissolved in degassed acetonitrile : water (9 : 1, 4.0 mL). The flask was wrapped in aluminum foil and then the chosen arene (furan, thiophene, 10–25 equiv.) added. The stock solution was thus charged in a “sunflow” reactor that was in turn ­positioned outdoor and exposed to sunlight for one hour. Biaryls 18a-­e were ­isolated from the raw photolysate by solvent removal followed by column chromatography. Experimental setup: The “sunflow” reactor (Figure 2.14) was flushed with argon and the stock solution was injected via a Luer–Lock syringe. The reactor was positioned outdoor and exposed to sunlight for one hour. X G

18a, X = O, G = 4-CH3CO, 82% b, X = O, G = 4-Br, 69% c, X = O, G = 2-CN, 33% d, X = S, G = 4-CN, 56% e, X = S, G = 4-CH3, 48%

X

(10–25 equiv.) MeCN-H2O, 1 h N2SO2CH3

G

via:

G

17a-e (0.05 M)

Syringe pump

FEP – tube OD = 1.6 mm ID = 1.0 mm

Length = 10 or 25 m

Figure 2.14  Photochemical flow reactor employed for the synthesis of 18a-­e. Source: Reproduced with permission from da Silva et al. [74]/John Wiley & Sons.

61

62

2  Introduction to Photochemistry for the Synthetic Chemist

2.3.4  Reactions Involving Singlet Oxygen The amount of Artemisinin, which is the most effective antimalarial drug, obtained via extraction from the plant sweet wormwood is obviously not sufficient to supply more than 300 million treatments needed each year, making the cost of such drug unsustainable. However, Artemisinin (20) can be prepared from naturally occurring dihydroartemisinic acid (19), and the key step in such synthesis is a Schenk-ene reaction with photo-­generated singlet oxygen (Scheme 2.7) formed upon photosensitization by using 1,9-­dicyanoanthracene (DCA, 0.5 mol%). In a continuous flow protocol, 19 (in turn recovered from the mother liquor accumulated in the commercial artemisinin extraction) is converted into 20 in up to 42% yield with a short residence time (1.5 minutes) and a calculated space-­time yield of 3500 kg m−3 day−1 [75, 76]. Typical procedure for the flow synthesis of 20: The photochemical reactor has been fed with a solution of 19 (0.5 M) and TFA in toluene, in the presence of 9,10-­dicyanoanthracene (DCA, 0.5 mol%) as the photosensitizer. An HPLC pump (Knauer, Smartline pump 100) has been used to deliver either substrate solution or pure solvent by means of a two-­way switch. In an ETFE T-­mixer (IDEX Health and Science), the solution was mixed with oxygen (flow rate, 5 mL min−1). The oxygen saturated solution was then pumped through the photoreactor that is equipped with a fluorinated ethylene–propylene copolymer (FEP) tubing and the plate cooled down to −20 °C with an immersion cooler. An LED module (420 nm emission wavelength) was employed for the irradiation. The solution was then re-­heated to 10 °C and then kept at room temperature. The collected solution was thus washed with aqueous saturated NaHCO3 and the solvent removed under vacuo to afford a yellow solid that was solubilized in acetonitrile and filtered through a PTFE syringe filter with 0.45 μm cutoff. Crystallization of the crude residue gave 20 in 42% yield (based on the starting 19).

2.3.5  Reactions Involving a Photocatalytic Step Photocatalysis has recently revolutionized the approach to organic synthesis through radical and radical ion intermediates. These fleeting species can be H CN

H DCA (0.5 mol%) H

H COOH

19 (0.5 M)

hv (420 nm) TFA (0.25 M) O2, Toluene

H

O O

O 20, 42%

Scheme 2.7  Photosensitized synthesis of Artemisinin 20.

DCA

CN

2.3  ­Selected Applications of Photochemical/Photocatalyzed Reaction

smoothly generated under extremely mild conditions through two main mechanisms, namely single electron transfer (SET) [23–30] or hydrogen atom transfer (HAT) [31–33]. As for the first mechanism, a seminal and interesting application is represented by the realization of [2+2] cycloadditions occurring via radical ionic intermediates to deliver cyclobutane rings. Indeed, this is a complementary strategy with respect to the above-­mentioned photosensitized [2+2] cycloaddition (see Scheme 2.4 above). Intriguingly, the exchange of an electron between the photocatalyst and the substrate, occurring upon irradiation, triggered the reactivity of two double bonds endowed with the same polarity (both nucleophilic or electrophilic). As an example, the crossed intermolecular [2+2] cycloaddition of acyclic enones has been reported to take place in the presence of the well-­known Ru(bpy)3 (bpy: 2,2′-­bipyridine) photoredox catalyst upon irradiation with visible light (Scheme 2.8a) [77]. A careful choice of the reaction partners was required, with the best performance being obtained by reacting an arylenone (e.g. 21a-­d) with an α,β-­unsaturated ketone lacking β-­substituents (e.g. the Michael acceptor 22). In the process, the arylenone was converted by the excited photocatalyst to the

(a) O

O

O +

Ar

23 W CFL

Me

Ru(bpy3)Cl2 (5 mol%) iPr2NEt (2 equiv.) 22 2.5 equiv. LiBF4 (4 equiv.) MeCN, 4 h, N2

Me 21a-d (0.1 M)

O via:

Ar

O

Ar

Me

Me 23a, Ar = C6H5, 84%, d.r. >10:1 b, Ar = 4-ClC6H4, 82%, d.r. >10:1 c, Ar = 4-MeOC6H4, 53%, d.r. >10:1 d, Ar = 2-furyl, 74%, d.r. >10:1

Me

(b)

Me

Me 20 W CFL

+ Me

OMe

25a,b 2 equiv.

24 (0.1 M) via:

Ru(bpm3)(BArF)2 MeO (0.25 mol%) MeCN, –15 °C, 1 h, air 26a, 4-Me, 86% b, 2-Me, 81%

Me

MeO Me

Scheme 2.8  Synthesis of four-­membered rings via photoredox catalyzed generation of (a) radical anions, (b) radical cations.

63

64

2  Introduction to Photochemistry for the Synthetic Chemist

corresponding radical anion that coupled with 22 ultimately delivering ­cyclobutanes 23a-­d. Interestingly, the reaction could be scaled up to the preparation of 1 g of product 23a upon irradiation with sunlight (only four hours exposition required, Figure 2.15). Typical procedure for the synthesis of 23a-­d: A dry 25 mL Schlenk tube was charged with the aryl enones 21a-­d (0.1 M), 22 (2.5 equiv.), Ru(bpy)3Cl2·6H2O (5 mol%), LiBF4 (4 equiv.), iPr2NEt (2 equiv.), and acetonitrile, and degassed by a freeze/pump/thaw cycle (3×) under nitrogen in the dark. The reaction was then allowed to stir and irradiated by a 23 W compact fluorescent lamp at a distance of 30 cm. Upon completion, the solvent was removed in vacuo and the residue purified by column chromatography on silica gel. Experimental setup: See Figure 2.15. By reversing the electron flow between the photocatalyst and the substrate, a smooth access to radical cationic intermediates is indeed possible. This strategy has been exploited to realize the crossed intermolecular [2+2] cycloaddition between styrenes (Scheme 2.8b) [78]. Thus, methoxy-­substituted styrene 24 and isomeric methylstyrenes 25a,b delivered cycloadducts 26a,b in the presence of the Ru(bpm)3 (bpm: 2,2′-­bipyrimidine) photocatalyst under visible light irradiation. In the process, the photocatalyst oxidized 24 to the corresponding radical cation, a key intermediate in the process. Furthermore, the choice of the photocatalyst was pivotal to obtain a good reaction yield, since it had to be competent in the mono-­electronic oxidation of 24, but not in that of 26, possibly causing an undesired cycloreversion event  [78]. In a follow-­up work, the importance of

Figure 2.15  Photochemical apparatus employed for the synthesis of 23a-­d. Source: Reprinted with permissions from Du and Yoon [77], Copyright © 2014 American Chemical Society.

2.3  ­Selected Applications of Photochemical/Photocatalyzed Reaction

radical chain processes in this type of [2+2] photocycloadditions occurring via radical ionic intermediates has been demonstrated thanks to quantum yield measurements [79]. Typical procedure for the synthesis of 26a,b: An oven-­dried 25 mL borosilicate test tube was charged with 24 (0.1 M), 25 (2 equiv.), and 0.25 mol% Ru(bpm)3(BArF)2. The test tube was cooled to −15 °C using a controlled-­temperature cooling bath and 6.7 mL of anhydrous CH2Cl2 was added. The reaction mixture was vigorously stirred under air in front of a 20 W CFL. The solvent was removed by rotary evaporation, and the residue was purified by flash-­column chromatography to afford the cycloadducts 26a,b. Experimental setup: The reaction vessel is a standard borosilicate test tube immersed in an acetone bath cooled to −15 °C using a Neslab immersion cooler. The light source is a 20 W GE Reveal CFL bulb installed in a clamp light. Turning to the HAT mechanism, one of the most investigated photocatalysts is undoubtedly the decatungstate anion [W10O32]4−, often used as the tetrabutylammonium salt (TBADT). This robust and versatile polyoxometalate derivative has been exploited to cleave the C─H bond in a library of hydrogen donors, including alkanes, ethers, amides and aldehydes, to generate the corresponding C-­centered radicals [31–33, 80]. As an example, the nucleophilic acyl radical obtained from heptaldehyde 27 has been conveniently exploited in the hydroalkylation of electron-­poor olefins (e.g. esters 28a,b) via a conjugate radical addition mechanism (Scheme 2.9). Notably, these processes have been carried out in a glass vessel upon solar light irradiation, in an example of the so-­called window-­ledge chemistry (Figure  2.16)  [55]. The desired products 29a,b were indeed obtained on up to ca. 10 g scale by merely exposing the reaction mixtures to sunlight for a few days without any need of artificial energy input. The versatility of decatungstate photocatalysis is further demonstrated by the possibility to develop a flow-­version of this chemistry. Thus, as previously O O H 27 (0.2 M)

+

OMe R 28a,b 1.25 equiv.

R

Direct sunlight (nBu4N)4[W10O32] (1 mol%) MeCN, 5–8 days, N2

via:

O

COOMe

29a (R = H), 70% b (R = COOMe), 90%

O C 6 H 13

Scheme 2.9  Sunlight-­induced formation of unsymmetrical ketones from aldehydes.

65

66

2  Introduction to Photochemistry for the Synthetic Chemist

Figure 2.16  Photochemical apparatus employed for the synthesis of 29a-­b.

described  [58], a photochemical reactor may be assembled by flowing the reaction mixture in an UV-­transparent FEP (fluorinated ethylene propylene) tubing wrapped around a medium pressure Hg-­vapors lamp (Figure 2.17). Thanks to this experimental setup, it was possible to synthesize the same unsymmetrical ketones obtained upon solar light irradiation (e.g. product 29b was prepared in 79% yield), albeit with a reduced irradiation time, thus increasing the overall productivity [81].

Figure 2.17  Photochemical flow reactor employed for the synthesis of 29a,­b. Source: Reprinted with permission from Bonassi et al. [81], Copyright © 2016 John Wiley & Sons, Inc.

Typical procedure for the synthesis of 29a,b (batch): A solution (100 mL) of heptaldehyde 27 (0.2 M), and 28a,b (1.25 equiv.) in the presence of TBADT (1 mol%) in MeCN was poured into a glass Pyrex vessel and purged for 10 minutes with nitrogen, capped and exposed to sunlight on a window ledge. The solvent was removed in vacuo from the photolyzed solution and 29a,b isolated by bulb-­to-­ bulb distillation. Experimental setup: Photocatalyzed reactions were carried out on solutions corresponding to a layer of about 1 cm thick with a surface area of 1 dm2 in the cylindrical vessel used (Figure  2.16, right side). Vessels of on the left side were used for small-­scale experiments for the optimization of the reaction conditions (left side).

 ­Reference

Typical procedure for the synthesis of 29a,b (flow): 27 and 28a,b in the presence of a catalytic amount of TBADT were dissolved in 50 mL of acetonitrile. The solution was charged in a flask and pumped through the flow apparatus (Figure 2.17); 29a,b were isolated upon column chromatography. Experimental setup: The photochemical reactor (Figure 2.17) was equipped with a water-­cooled 500 W medium-­pressure mercury lamp. All the tubings were made of fluorinated ethylene propylene (FEP, outer diameter: 3.18 mm; inner diameter: 2.1 mm; reactor volume = 50 mL).

2.4 ­Conclusions Summing up, it is not difficult work to arrange a small photochemical lab to explore the viability of photochemical steps. Lamps operating with an emission centered at 254, 305, and 350 nm in the UV region and LEDs for the visible spectrum are often sufficient for this purpose. As reactor vessels, common glassware (round-­bottom flask, vials, etc.) and some quartz tubes are well fitted for this aim. In case of sunlight-­promoted reactions, the use of this freely available light source does not require any investment. Exhaustive textbooks and reviews dealing with the basis of photochemistry [82–84], as well as dealing with the synthetic aspects of organic photochemistry [85–87], are available. For references on the use of photochemistry/photocatalysis for undergraduate organic chemistry laboratory experiments, see Refs. [88–90].

­Acknowledgment We thank Miss Alessandra Del Tito for taking some pictures included in this chapter.

­References 1 Albini, A. and Fagnoni, M. (2004). Green chemistry and photochemistry were born at the same time. Green Chemistry 6: 1–6. 2 Albini, A. and Fagnoni, M. (2008). 1908: Giacomo Ciamician and the concept of green chemistry. ChemSusChem 1: 63–66. 3 Hoffmann, N. (2012). Photochemical reactions of aromatic compounds and the concept of the photon as a traceless reagent. Photochemical & Photobiological Sciences 11: 1613–1641.

67

68

2  Introduction to Photochemistry for the Synthetic Chemist

4 Bonfield, H.E., Knauber, T., Lévesque, F. et al. (2020). Photons as a 21st century reagent. Nature Communications 11: 804. 5 Protti, S., Ravelli, D., and Fagnoni, M. (2019). Wavelength-­dependence and wavelength-­selectivity in photochemical reactions. Photochemical & Photobiological Sciences 18: 2094–2101. 6 Balzani, V., Ceroni, P., and Juris, A. (2014). Photochemistry and Photophysics: Concepts, Research, Applications. John Wiley & Sons, Inc. 7 Noyori, P. (2010). Insight: Green chemistry: the key to our future. Tetrahedron 66: 1028. 8 White, J.D. (1998). George Buchi. Obituary. Organic Syntheses 77: xxiii–xxvi. 9 Chang, L., An, Q., Duan, L. et al. (2022). Alkoxy radicals see the light: new paradigms of photochemical synthesis. Chemical Reviews 122: 2429–2486. 10 Guo, J.-­J., Hu, A., and Zuo, Z. (2018). Photocatalytic alkoxy radical-­mediated transformations. Tetrahedron Letters 59: 2103–2111. 11 Kamijo, S. (2019). Ketones and aldehydes. In: Photoorganocatalysis in Organic Synthesis (ed. M. Fagnoni, S. Protti, and D. Ravelli), 1–38. Singapore: World Scientific Publishing Europe Ltd. 12 D’Auria, M. (2019). The Paternò–Büchi reaction – a comprehensive review. Photochemical & Photobiological Sciences 18: 2297–2362. 13 Albini, A. and Fagnoni, M. (2013). Photochemically-­Generated Intermediates in Synthesis. Hoboken: John Wiley & Sons. 14 Crespi, S. and Fagnoni, M. (2020). Generation of alkyl radicals: from the Tyranny of Tin to the photon democracy. Chemical Reviews 120: 9790–9833. 15 Ravelli, D., Protti, S., and Fagnoni, M. (2016). Carbon-­carbon bond forming reactions via photogenerated intermediates. Chemical Reviews 116: 9850–9913. 16 Shields, D.J., Chakraborty, M., Abdelaziz, N. et al. (2020). Review of laser flash photolysis of organic molecules (2015–2018). Photochemistry 48: 70–121. 17 Protti, S., Ravelli, D., and Fagnoni, M. (2022). Designing radical chemistry by visible-­light promoted homolysis. Trends in Chemistry 4: 305–317. 18 Dichiarante, V., Protti, S., and Fagnoni, M. (2017). Phenyl cation: a versatile intermediate. Journal of Photochemistry and Photobiology A: Chemistry 339: 103–113. 19 Großkopf, J., Kratz, T., Rigotti, T. et al. (2022). Enantioselective photochemical reactions enabled by triplet energy transfer. Chemical Reviews 122: 1626–1653. 20 Pipiri, I., Buscemi, S., Piccionciello, A.P. et al. (2018). Photochemically produced singlet oxygen: applications and perspectives. ChemPhotoChem 2: 535–547. 21 You, Y. (2018). Chemical tools for the generation and detection of singlet oxygen. Organic & Biomolecular Chemistry 16: 4044–4060. 22 Wau, J.S., Robertson, M.J., and Oelgemöller, M. (2021). Solar photooxygenations for the manufacturing of fine chemicals—­technologies and applications. Molecules 26: 1685.

 ­Reference

23 Narayanam, J.M.R. and Stephenson, C.R.J. (2011). Visible light photoredox catalysis: applications in organic synthesis. Chemical Society Reviews 40: 102–113. 24 Xuan, J. and Xiao, W.-­J. (2012). Visible-­light photoredox catalysis. Angewandte Chemie, International Edition 51: 6828–6838. 25 Reckenthäler, M. and Griesbeck, A.G. (2013). Photoredox catalysis for organic syntheses. Advanced Synthesis & Catalysis 355: 2727–2744. 26 Xi, Y., Yi, H., and Lei, A. (2013). Synthetic applications of photoredox catalysis with visible light. Organic & Biomolecular Chemistry 11: 2387–2403. 27 Stephenson, C.R.J., Yoon, T.P., and MacMillan, D.W.C. (ed.) (2018). Visible Light Photocatalysis in Organic Chemistry. Weinheim, Germany: Wiley-­VCH. 28 Hossain, A., Bhattacharyya, A., and Reiser, O. (2019). Copper’s rapid ascent in visible-­light photoredox catalysis. Science 364: eaav9713. 29 König, B. (ed.) (2020). Chemical Photocatalysis, 2ee. De Gruyter. 30 Pitre, S.P., Overman, L.E. (2022). Strategic use of visible-­light photoredox catalysis in natural product synthesis. Chemical Reviews 122: 1717–1751. 31 Protti, S., Ravelli, D., and Fagnoni, M. (2015). Photocatalytic C-­H activation via hydrogen atom transfer in synthesis. ChemCatChem 7: 1516–1523. 32 Cao, H., Tang, X., Tang, H. et al. (2021). Photoinduced intermolecular hydrogen atom transfer reactions in organic synthesis. Chemical Catalysis 1: 523–598. 33 Capaldo, L., Ravelli, D., and Fagnoni, M. (2022). Direct photocatalyzed hydrogen atom transfer (HAT) for aliphatic C–H bonds elaboration. Chemical Reviews 122: 1875–1924. 34 Fukuzumi, S. and Ohkubo, K. (2014). Organic synthetic transformations using organic dyes as photoredox catalysts. Organic & Biomolecular Chemistry 12: 6059–6071. 35 Romero, N.A. and Nicewicz, D.A. (2016). Organic photoredox catalysis. Chemical Reviews 116: 10075–10166. 36 Sideri, I.K., Voutyritsa, E., and Kokotos, C.G. (2018). Photoorganocatalysis, small organic molecules and light in the service of organic synthesis: the awakening of a sleeping giant. Organic & Biomolecular Chemistry 16: 4596–4614. 37 Fagnoni, M., Protti, S., and Ravelli, D. (ed.) (2019). Photoorganocatalysis in Organic Synthesis. Singapore: World Scientific Publishing Europe Ltd. 38 Albini, A. and Fagnoni, M. (ed.) (2010). Handbook of Synthetic Photochemistry. Weinheim: Wiley-­VCH. 39 Ramamurthy, V. and Venkatesan, K. (1987). Photochemical reactions of organic crystals. Chemical Reviews 87: 433–481. 40 Garcia-­Garibay, M.A. (2007). Molecular crystals on the move: from single-­crystal-­ to-­single-­crystal photoreactions to molecular machinery. Angewandte Chemie, International Edition 46: 8945–8947. 41 Kuzmanich, G., Natarajan, A., Chin, K.K. et al. (2008). Solid state photodecarbonylation of diphenylcyclopropenone: a quantum chain process

69

70

2  Introduction to Photochemistry for the Synthetic Chemist

made possible by ultrafast energy transfer. Journal of the American Chemical Society 130: 1140–1141. 42 Knowles, J.P., Elliott, L.D., and Booker-­Milburn, K.I. (2012). Flow photochemistry: old light through new windows. Beilstein Journal of Organic Chemistry 8: 2025–2052. 43 Garlets, Z.J., Nguyen, J.D., and Stephenson, C.R.J. (2014). The development of visible-­light photoredox catalysis in flow. Israel Journal of Chemistry 54: 351–360. 44 Schuster, E.M. and Wipf, P. (2014). Photochemical flow reactions. Israel Journal of Chemistry 54: 361–370. 45 Buglioni, L., Raymenants, F., Slattery, A. et al. (2022). Technological innovations in photochemistry for organic synthesis: flow chemistry, high-­throughput experimentation, scale-­up, and photoelectrochemistry. Chemical Reviews 122: 2752–2906. 46 Matsushita, Y., Ichimura, T., Ohba, N. et al. (2007). Recent progress on photoreactions in microreactors. Pure and Applied Chemistry 79: 1959–1968. 47 Roberge, D.M., Zimmermann, B., Rainone, F. et al. (2008). Microreactor technology and continuous processes in the fine chemical and pharmaceutical industry: Is the revolution underway? Organic Process Research & Development 12: 905–910. 48 Coyle, E.E. and Oelgemöller, M. (2008). Micro-­photochemistry: photochemistry in microstructured reactors. The new photochemistry of the future? Photochemical & Photobiological Sciences 7: 1313–1322. 49 Basso, A. and Capurro, P. (2021). Recent applications of photochemistry on large-­scale synthesis (2015–2019). Photochemistry 48: 294–322. 50 Candish, L., Collins, K.D., Cook, G.C. et al. (2022). Photocatalysis in the Life Science Industry. Chemical Reviews 122: 2907–2980. 51 As a representative supplier for UV lamps. https://www.heliosquartz.com/ (accessed 11 April 2023). 52 For the applications of Kessil lamps in synthesis. https://kessil.com/products/ science_main.php (accessed 11 April 2023). 53 Ravelli, D., Protti, S., and Fagnoni, M. (2016). Application of visible and solar light in organic synthesis. In: Applied Photochemistry: When Light Meets Molecules (ed. G. Bergamini and S. Silvi), 281–342. Switzerland: Springer International Publishing. 54 Schultz, D.M. and Yoon, T.P. (2014). Solar synthesis: prospects in visible light photocatalysis. Science 343: 1239176. 55 Ravelli, D., Protti, S., and Fagnoni, M. (2016). Decatungstate anion for photocatalyzed “Window Ledge” reactions. Account of Chemical Research 49: 2232–2242. 56 For the characteristics of a SolarBox. https://cofomegra.it/en/solarbox-­1500-­3000/ (accessed 11 April 2023).

 ­Reference

57 Protti, S. and Fagnoni, M. (2009). The sunny side of chemistry: green synthesis by solar light. Photochemical & Photobiological Sciences 8: 1499–1516. 58 Hook, B.D.A., Dohle, W., Hirst, P.R. et al. (2005). Practical flow reactor for continuous organic photochemistry. The Journal of Organic Chemistry 70: 7558–7564. 59 Rehm, T.H. (2020). Reactor technology concepts for flow photochemistry. ChemPhotoChem 4: 235–254. 60 For the different types of photochemical flow reactors. https://www.corning.com/ worldwide/en/innovation/corning-­emerging-­innovations/advanced-­flow-­ reactors.html (accessed 11 April 2023) 61 For continuous flow devices. https://www.syrris.com/product/asia-­ photochemistry-­system/ (accessed 11 April 2023) 62 For the design of a falling film reactor. https://photoreactors.com/products/ falling-­film-­photoreactors/ (accessed 11 April 2023). 63 Arias-­Rotondo, D.M. and Mc Cusker, J.K. (2016). The photophysics of photoredox catalysis: a roadmap for catalyst design. Chemical Society Reviews 45: 5803–5820. 64 Buzzetti, L., Crisenza, G.E.M., and Melchiorre, P. (2019). Mechanistic studies in photocatalysis. Angewandte Chemie, International Edition 58: 3730–3747. 65 von Lintig, J., Kiser, P.D., Golczak, M. et al. (2010). The biochemical and structural basis for trans-­to-­cis isomerization of retinoids in the chemistry of vision. Trends in Biochemical Sciences 35: 400–410. 66 Huang, M., Zhang, L., Pan, T. et al. (2022). Deracemization through photochemical E/Z isomerization of enamines. Science 375: 869–874. 67 Folkmann, M. P. and Hansen, E. T. (2007). Isomerisation of pharmaceutical intermediates. EP 1844010. 68 Rigotti, T., Schwinger, D.P., Graßl, R. et al. (2022). Enantioselective crossed intramolecular [2+2] photocycloaddition reactions mediated by a chiral chelating Lewis acid. Chemical Science 13: 2378–2384. 69 Mateos, J., Vega-­Peñaloza, A., Franceschi, P. et al. (2020). A visible-­light Paternò– Büchi dearomatisation process towards the construction of oxeto-­indolinic polycycles. Chemical Science 11: 6532–6538. 70 Govaerts, S., Angelini, L., Hampton, C. et al. (2020). Photoinduced olefin diamination with alkylamines. Angewandte Chemie International Edition 59: 15021–15028. 71 Steiner, A., de Frutos, O., Rincón, J.A. et al. (2021). N-­Chloroamines as substrates for metal-­free photochemical atom-­transfer radical addition reactions in continuous flow. Reaction Chemistry & Engineering 6: 2434–2441. 72 Qiu, D., Lian, C., Mao, J. et al. (2020). Dyedauxiliary groups, an emerging approach in organic chemistry. The case of arylazo sulfones. The Journal of Organic Chemistry 85: 12813–12822. 73 Fagnoni, M. (2022). Colored compounds for eco-­sustainable visible-­light promoted syntheses. In: Sustainable Organic Synthesis: Tools and Strategies (ed. S. Protti and A. Palmieri), 150–180. The Royal Society of Chemistry.

71

72

2  Introduction to Photochemistry for the Synthetic Chemist

74 da Silva Jr, P.E., Amin, H.I.M., Nauth, A.M. et al. (2018). Flow photochemistry of azosulfones: application of “Sunflow” reactors. ChemPhotoChem 2: 878–883. 75 Kopetzki, D., Lévesque, F., and Seeberger, P.H. (2013). A continuous-­flow process for the synthesis of artemisinin. Chemistry A European Journal 19: 5450–5456. 76 Lévesque, F. and Seeberger, P.H. (2012). Continuous-­flow synthesis of the anti-­ malaria drug artemisinin. Angewandte Chemie, International Edition 51: 1706–1709. 77 Du, J. and Yoon, T.P. (2009). Crossed intermolecular [2+2] cycloadditions of acyclic enones via visible light photocatalysis. Journal of the American Chemical Society 131: 14604–14605. 78 Ischay, M.A., Ament, M.S., and Yoon, T.P. (2012). Crossed intermolecular [2+2] cycloaddition of styrenes by visible light photocatalysis. Chemical Science 3: 2807–2811. 79 Cismesia, M.A. and Yoon, T.P. (2015). Characterizing chain processes in visible light photoredox catalysis. Chemical Science 6: 5426–5434. 80 Protti, S., Ravelli, D., Fagnoni, M. et al. (2009). Solar light-­driven photocatalyzed alkylations. Chemistry on the window ledge. Chemical Communications 47: 7351–7353. 81 Bonassi, F., Ravelli, D., Protti, S. et al. (2015). Decatungstate photocatalyzed acylations and alkylations in flow via hydrogen atom transfer. Advanced Synthesis & Catalysis 357: 3687–3695. 82 Horspool, W.H. and Armesto, D. (1992). Organic Photochemistry: A Comprehensive Treatment. London: Ellis Horwood. 83 Turro, N.J., Ramamurthy, V., and Scaiano, J.C. (2009). Modern Molecular Photochemistry of Organic Molecules. Basingstoke: Palgrave Macmillan. 84 Klan, P. and Wirz, J. (2009). Photochemistry of Organic Compounds. Wiley-­Blackwell. 85 Mattay, J. and Griesbeck, A. (ed.) (2008). Photochemical Key Steps in Organic Synthesis: An Experimental Course Book. Wiley Weinheim. 86 Griesbeck, A., Oelgemöller, M., and Ghetti, F. (ed.) (2012). Handbook of Organic Photochemistry and Photobiology, 3ee. CRC: Press. 87 Bergamini, G. and Silvi, S. (ed.) (2016). Applied Photochemistry: When Light Meets Molecules. Switzerland: Springer International Publishing. 88 Contreras-­Cruz, D.A., Cantû-­Reyes, M., García-­Sánchez, J.M. et al. (2019). Shedding Blue Light on the Undergraduate Laboratory: an easy-­to-­assemble LED photoreactor for aromatization of a 1,4-­dihydropyridine. Journal of Chemical Education 96: 2015–2020. 89 Chen, S. (2018). let there be light: hypothesis-­driven investigation of ligand effects in photoredox catalysis for the Undergraduate Organic Chemistry Laboratory. Journal of Chemical Education 95: 872–875. 90 Cruz, C.L., Holmberg-­Douglas, N., Onuska, N.P.R. et al. (2020). Development of a large-­enrollment course-­based research experience in an Undergraduate Organic Chemistry Laboratory: structure–function relationships in pyrylium photoredox catalysts. Journal of Chemical Education 97: 1572–1578.

73

3 How to Confidently Become an Electrosynthetic Practitioner Sylvain Charvet1,*, Taline Kerackian1,*, Camille Z. Rubel1,2,*, and Julien C. Vantourout1 1 Institut de Chimie et Biochimie Moléculaires et Supramoléculaires (ICMBS, UMR 5246 du CNRS), Université Lyon, Villeurbanne, France 2 Department of Chemistry, The Scripps Research Institute, La Jolla, CA, USA

Glossary Current   Electric current is defined as the flow of electric charge through a conductive material or medium. It is the rate of movement of electric charge in a specific direction, usually measured in amperes (A). The electric charge can be carried by electrons in a metallic conductor, ions in an electrolyte, or any other charged particles in a plasma. The direction of the electric current is defined as the direction of the flow of positive charge, even though in most cases it is the negative charges that move. Electric current plays a fundamental role in a wide range of electrical and electronic devices, including motors, generators, batteries, and power distribution systems. Current density  Current density is defined as the amount of electric current flowing through a unit area perpendicular to the direction of the current flow. It is a vector quantity that describes the magnitude and direction of the flow of electric charge in a material. Mathematically, it can be represented by the symbol J and is defined as J = I/A, where J is the current density, I is the current flowing through the material, and A is the surface area of the material through which the current is flowing. The unit of current density is amperes per square meter (A.m−2) in the SI system of units. Current density is an important parameter in many areas of electrical engineering, such as the design of electrical conductors, semiconductors, and electronic devices. * These authors contributed equally. Enabling Tools and Techniques for Organic Synthesis: A Practical Guide to Experimentation, Automation, and Computation, First Edition. Edited by Stephen G. Newman. © 2023 John Wiley & Sons, Inc. Published 2023 by John Wiley & Sons, Inc.

74

3  How to Confidently Become an Electrosynthetic Practitioner

It is also used to analyze and optimize the performance of electrical circuits and systems. Cyclic voltammetry  Cyclic voltammetry (CV) is a widely used electrochemical technique. During a typical CV experiment, the voltage is ramped up and down in a repetitive cycle, while the resulting current response is recorded. This technique is commonly used to study the electrochemical behavior of a wide range of chemical and biological systems. Some uses include the determination of the redox potentials of redox-­active species, the kinetics of electron transfer reactions, and the characterization of electrode surfaces and electroactive molecules. CV data can also provide a wealth of information about the electrochemical system being studied, including the nature and concentration of the electroactive species present, the number and types of electron transfer reactions occurring, and the rate of these reactions. Electrochemistry  Electrochemistry is the study of electron movement in an oxidation or reduction reaction at a polarized electrode surface. This term can an also refer to the transfer of electrons from one chemical species to another. Electrolyte  An electrolyte is a substance that contains ions and can conduct electricity. When an electrolyte is placed in a solution, the ions in the solution become mobile and can carry an electrical current through the solution. Electrolytes can be either liquids or solids, and they can be either organic or inorganic compounds. In aqueous solutions, common examples of electrolytes include salts such as sodium chloride (NaCl), potassium chloride (KCl), and calcium chloride (CaCl2). In organic solvents, common examples of electrolytes include salts such as tetrabutylammonium tetrafluoroborate (TBABF4), potassium hexafluorophosphate (KPF6), and lithium bromide (LiBr). Electrolytes are important in electrochemical reactions because they play a crucial role in the transfer of electrons and the conduction of electrical charge. They are often used in batteries, fuel cells, and other electrochemical devices. Electrosynthesis  Electrosynthesis refers to the process of using electrical energy to drive a chemical reaction, typically involving the conversion of a molecule into a different molecule. This process involves the application of an electric potential or current to a reactant solution, causing electrons to be transferred between the electrode and the solution at the electrode surface, which in turn drives the desired chemical reaction. Electrosynthesis has been demonstrated in the production of various substances, such as pharmaceuticals, organic compounds, and electroplating of metals. Faraday   The Faraday is defined as the amount of electrical charge (in coulombs) that is transferred when one mole of electron flows through a conductor during an electrochemical reaction. The term “Faraday” is used to describe the amount of electrical charge transferred during an electrochemical reaction. It is named after the English scientist Michael Faraday who made significant contributions to the field of electromagnetism and electrochemistry in the

Glossary  75

nineteenth century. It is often used in the form of the Faraday constant, which is approximately equal to 96 485 coulombs per mole of electrons. The Faraday is an important unit of measurement in electrochemistry as it is used to calculate the amount of substance produced or consumed during an electrochemical reaction. It is also used to determine the efficiency of electrochemical processes and to calculate the amount of electric current needed to produce a certain amount of substance in an electrolytic cell. Metallophotoredox  Metallophotoredox is a type of catalytic reaction that involves the use of a photosensitizer and a metal catalyst to facilitate electron transfer reactions. The photosensitizer is typically a molecule that absorbs light and becomes electronically excited, while the metal catalyst serves to mediate electron transfer reactions. Together, they enable a wide range of organic transformations, including cross-­coupling reactions, hydrogenation reactions, and oxidation reactions. Metallophotoredox has become an increasingly important area of research in synthetic chemistry and is a powerful tool for the efficient construction of complex molecules with high levels of control and selectivity. Oxidation and anode  An electrochemical oxidation is a chemical reaction that involves the loss of electrons by a substance through the application of an electric current. In this process, the species is subjected to an electric potential, which causes it to undergo an oxidation reaction, resulting in the formation of an oxidized product. During an electrochemical oxidation, the oxidation half-­reaction takes place at the anode, which is the electrode connected to the positive terminal of the power source. The anode attracts negatively charged ions (i.e. anions) from the solution, and these anions are oxidized by the loss of electrons. Electrochemical oxidation is used to convert a wide range of organic and inorganic substances into their oxidized forms, which can have a variety of applications in industry, environmental remediation, and energy storage. Passivation  Electrochemical passivation is the process by which a metal surface is made more resistant to corrosion through the formation of a thin, protective oxide layer. This oxide layer is created through an electrochemical reaction between the metal and an electrolyte, such as oxygen or water. The process typically occurs by applying a potential difference between the metal and the electrolyte, which causes the metal to oxidize and form the protective oxide layer. Once the oxide layer has formed, it acts as a barrier to further oxidation and corrosion, effectively “passivating” the metal surface. In the context of electrosynthesis, the process hampers the desired chemical reaction from occurring. Potential  Potential is the potential energy difference between two points in an electrochemical system due to the presence of charged particles or ions, which can affect the direction and rate of chemical reactions. It is expressed in units of volts (V) and is an important concept in fields such as electrochemistry, biology, and physics. Current and potential are linked by

76

3  How to Confidently Become an Electrosynthetic Practitioner

Ohm’s law. This law states that the current through a conductor between two points is directly proportional to the voltage across the two points, provided that the temperature and other physical conditions of the conductor remain constant. In other words, as the voltage increases, the current through the circuit will also increase proportionally. Mathematically, Ohm’s law can be expressed as: I = V/R, where I is the current flowing through the conductor, V is the voltage across the conductor, and R is the resistance of the conductor. Reduction and cathode  An electrochemical reduction is a chemical reaction that involves the gain of electrons by a substance through the application of an electric current. In this process, the species is subjected to an electric potential, which causes it to undergo a reduction reaction, resulting in the formation of a reduced product. During an electrochemical reduction, the reduction half-­ reaction takes place at the cathode, which is the electrode connected to the negative terminal of the power source. The cathode attracts positively charged ions (i.e. cations) from the solution, and these cations are reduced by the gain of electrons. Electrochemical reduction is commonly used in electroplating, in which a metal is reduced onto a substrate by electrochemical means. It is also used in the production of chemicals and fuels, such as hydrogen gas and ammonia, by electrolysis of water and nitrogen gas, respectively. Synthon  In organic chemistry, a synthon is a hypothetical or real molecular fragment (or intermediate) that can be used as a building block in the synthesis of more complex molecules. An organic synthon is a synthetic equivalent of a functional group or group of atoms, which can be used to prepare a target molecule via a series of chemical reactions. Organic synthons are typically used in retrosynthetic analysis, which is a strategy for planning organic syntheses by breaking down a target molecule into simpler precursors. The use of synthons can simplify retrosynthetic analysis by reducing a complex molecule to a smaller and more easily handled fragment.

Abbreviations Al C DMA DMF DMSO ERCC Fe FTO GC-­MS HCl

Aluminum Carbon N,N-­Dimethyl acetamide N,N-­Dimethyl formamide Dimethyl sulfoxide Electrophile reductive cross-­coupling Iron Fluorine-­doped tin oxide. Gas chromatography – mass spectrometry Hydrochloric acid

3.1 ­Introductio

LC-­MS LiBr Mg Ni NMR Pt Rpm RVC THF TLC Zn

Liquid chromatography – mass spectrometry Lithium bromide Magnesium Nickel Nuclear magnetic resonance Platinum Rotations per minute Reticulated vitreous carbon Tetrahydrofuran Thin layer chromatography Zinc

3.1 ­Introduction Electrosynthesis is the process of using electricity to induce chemical reactions. The history of organic electrosynthesis started almost 200 years and has led to an extensive breadth of research since 2000 [1]. Electrosynthesis has a wide range of applications in various industries, including chemical synthesis, energy storage, environmental remediation, and pharmaceuticals [2–7]. Electrosynthesis can also enable the production of chemicals and materials that are difficult or impossible to obtain through traditional chemical processes. In the chemical industry, electrosynthesis has the potential to transform the production of specialty chemicals and intermediates such as those in cosmetics, agrochemicals, and pharmaceuticals [2, 3]. Traditional chemical synthesis methods for these compounds are often energy-­intensive and result in significant waste. Electrosynthesis can provide a more sustainable alternative, enabling the production of these compounds with high selectivity and purity. In the energy storage sector, electrosynthesis is being explored as a means of converting renewable energy sources, such as solar and wind power, into chemical energy [4, 5]. This can be achieved through the production of fuels, such as hydrogen, from water using electrosynthesis  [6, 7]. Electrosynthesis can also be used for environmental remediation, enabling the removal of pollutants from contaminated water and soil [8, 9]. However, electrolysis remains a very under-­used procedure for the preparation of organic compounds in both the laboratory and industry. This is because, historically, organic electrosynthesis publications were inaccessible to synthetic organic chemists. Seminal reports were not written with the typical synthetic chemist reader in mind, which led to multiple misunderstandings on why, when, and how one should use electrochemistry in a synthetic laboratory. For example, many electrochemical reactions are reported while performing analytic measurements such as impedance, but these are not necessarily useful at a synthetic scale. Unfortunately, most reports often fail to distinguish if a reported transformation

77

78

3  How to Confidently Become an Electrosynthetic Practitioner

is of practical use or not, leading to a body of literature that is misleading and confusing, especially to non-­electrochemists. In addition, too often an attempt to repeat an electrosynthesis in a different laboratory leads to lower selectivity and/ or yield due to insufficient detail in the description of the cell (geometry, dimensions, materials of components. . .) and the conditions (solvent, concentration of reagents and electrolyte, pH, cell current, temperature. . .). Thus, in this introduction, our goal is to clarify why it is important to use electrosynthesis and how to perform electrochemical reactions in the context of organic chemistry.

3.2 ­General Definition of Organic Electrosynthesis As previously mentioned, organic electrosynthesis is a branch of organic chemistry that uses electrochemistry to carry out chemical reactions involving organic compounds. In organic electrosynthesis, an electric current is passed through a solution containing the organic compounds, which causes the transfer of electrons between the reactants and the electrodes. This electron transfer results in the formation of radicals, ions, and other reactive intermediates, which can then undergo further chemical transformations to produce the desired products.

3.3 ­Why is Organic Electrosynthesis Used? There are several reasons why organic electrosynthesis is advantageous. Organic electrosynthesis is often a cleaner and more sustainable alternative to traditional chemical synthesis. Indeed, while chemical processes require the use of toxic reagents and generate large amounts of waste, electrochemical reactions are driven only by electrons. In addition, electrochemistry can be highly selective, allowing chemists to target specific chemical bonds in a molecule, which is of particular interest in the context of complex organic molecule synthesis. Moreover, many organic electrosynthesis reactions can be carried out at mild temperatures and pressures, which can reduce the risk of unwanted side reactions and improve overall reaction efficiency. Finally, organic electrosynthesis is a safer alternative to traditional chemical synthesis, as there is often less risk of fire, explosion, or other hazards associated with using highly oxidizing or reducing chemical reagents.

3.4 ­How is Organic Electrosynthesis Performed? General steps to perform electrosynthesis are listed below and will be elaborated on in Section 3.6 and 3.7: 1) Determine the reaction you want to carry out and identify the reactants and the product(s).

3.5  ­Where to Start with Electrosynthesis

2) Select the appropriate electrodes for your reaction. Electrodes are typically made of a conductive material such as platinum, graphite, or carbon. 3) Choose an appropriate electrolyte that can conduct electricity. The choice of electrolyte will depend on the nature of the reaction. 4) Setup an electrochemical cell by placing the two electrodes in the electrolyte solution. The two electrodes should be connected to a source of electrical energy, such as a battery or power supply. 5) Apply a voltage or a current to the cell to drive the reaction. The voltage or ­current required will depend on the nature of the reaction and the ­electrodes used. 6) Monitor the progress of the reaction by measuring the current passing through the cell, and the change in the concentrations of the reactants and products over time. All typical analytical techniques like TLC, GC, or HPLC can be used. 7) Once the reaction is complete, collect the product and purify it if necessary as you would do for an organic reaction.

3.5 ­Where to Start with Electrosynthesis? Over the years, many books and reviews have regularly highlighted the diversity of chemistry possible under electrochemical conditions leading to more sustainable and efficient processes. Key general references are listed below, which are excellent starting points for organic chemists to learn about both the practical and fundamental aspects of using electrochemistry for synthesis:

Selected General Reviews Baran: Synthetic Organic Electrochemical Methods Since 2000: On the Verge of a Renaissance [1] Nagaki: Modern Strategies in Electroorganic Synthesis [10] Fronatana-­Uriba: Organic electrosynthesis: a promising green methodology in organic chemistry [11] Little: Redox catalysis in organic electrosynthesis: basic principles and recent developments [12] Pletcher: Organic Electrosynthesis [13] Zeng: Use of Electrochemistry in the Synthesis of Heterocyclic Structures [14] Waldvogel: Electrochemical Screening for Electroorganic Synthesis [15]

Selected General Guides Hilt: Basic Strategies and Types of Applications in Organic Electrochemistry [16] Baran: A Survival Guide for the “Electro-­curious” [17] Lam: A practical guide to electrosynthesis [18]

79

80

3  How to Confidently Become an Electrosynthetic Practitioner

It is important to note that electrosynthesis can be a complex process and may require optimization of various parameters such as electrode materials, ­electrolyte, voltage, and current/potential. Careful control of these parameters can lead to increased product yield, purity, and selectivity. In addition, being able to ­confidently reproduce the results is also an important factor as it will favor the uptake of electrochemical reactions. Challenges associated with optimization, parameter control, and reproducibility have been substantially improved by the development of standardized equipment such as the ElectraSyn 2.0 from IKA. Since its commercialization in 2017, the field of electrochemistry has ­witnessed an increase of interest due to the practicality offered by the instrument. Given the breadth of recent reviews, we have chosen to make this book chapter a focused and practical “how-­to” guide for beginners to the field. The first part of this chapter will thus aim to teach someone who has never conducted an electrochemical experiment, or who has never used an IKA ElectraSyn 2.0, how to do so confidently. It contains pictures and explanations, as well as tips and tricks. It focuses on the laboratory aspect of electrochemical experimentation rather than the conceptual aspect of experiment design, which will be covered in the case study section. The IKA ElectraSyn 2.0 is relatively inexpensive (MSRP US $2,650 for the most basic model at the time of writing) and is an excellent starting point. However, designing custom electrochemical reactors is not particularly difficult, even for beginners. The following discussion on the ElectraSyn 2.0  will still be informative for those designing their own systems thanks to the descriptions of useful parameters including current, potential, electrolytes, and electrodes. Readers are also encouraged to read the reviews by the Hilt [16], Baran [17], and Lam [18] research groups for beginner-­friendly discussion on designing custom reactors.

3.6 ­Electrasyn 2.0 3.6.1  Machine and Consumables 3.6.1.1  Opening the IKA ElectraSyn 2.0 Box

When you buy an IKA ElectraSyn 2.0, you receive a briefcase-­like black plastic case with various components (Figure 3.1a). This case contains the ElectraSyn, a 20 mL glass vial and its white cap, a square carrier for connecting vials to the ElectraSyn, the plug-­in power and various adapters, and a stir plate. The ElectraSyn can support various attachments including a stir plate, the square carrier used to connect one ElectraSyn vial, or the carousel used to connect six vials at once (Figure 3.1b). One can also attach the Gogo module (Figure 3.1b), which allows you to connect the vial to the ElectraSyn but place it elsewhere for heating, sonication, or cooling. The carousel and the Gogo module are not included in the ElectraSyn 2.0 basic package and are considered accessories.

3.6 ­Electrasyn 2.

(a) Stir plate

Plug-in power

Square carrier Electrasyn 2.0

Vial with cap

(b)

Stir plate

Square carrier

Carousel

Gogo module

Figure 3.1  Commercially available ElectraSyn 2.0. (a) Commercial box; (b) Equipment.

3.6.1.2  Cell (Vial and Cap)

The reaction cell consists of the ElectraSyn vial and cap. 1) Glassware Glassware comes in 1, 2, 5, 10, and 20 mL sizes. These can be purchased on the IKA website or can be made by a glassblower (Figure 3.2). 2) Caps and septum You can purchase vials with or without caps. The caps come with a smaller screw-­on cap with a septum for introducing reagents into reactions. However, we have found the seal of these to be poor, so if air-­sensitivity is an issue, it is recommended to instead cover this opening with a small rubber septum, which can be wrapped in parafilm or tape for even better sealing (Figure 3.2). With tape

Vials of 2, 5, and 20 mL

Figure 3.2  Commercially available accessories.

With parafilm

Caps for vial

Normal cap

81

82

3  How to Confidently Become an Electrosynthetic Practitioner

3.6.1.3  Electrodes

Each electrochemical reaction must be run with two electrodes, a cathode and an anode. These come in many varieties. A third electrode can also be used as a reference electrode to properly control the potential but is not mandatory, especially for reactions run under constant current [16, 18]. 1) Cathode and anode There are various types of electrodes. Any (semi)conductive material can serve as an electrode material, ranging from cheap aluminum foil to expensive platinum plates or boron-­doped diamond electrodes. Knowing the reaction occurs at the surface of electrodes, the choice of the material could have a huge impact on the reactivity. The best way to find the right electrodes is typically trial-­and-­ error. To start, carbon-­based materials are most of the time used due to the broad variety of surfaces available including graphite, carbon felt, glassy carbon, and reticulated vitreous carbon (RVC) (Figure 3.3). The choice will therefore depend on the type of electrochemical reactions ( for more details on this particular topic the reader should consider reading Refs. [17, 18]). For example, for a purely reductive process, a sacrificial anode needs to be used (that means the anode will be consumed during the reaction). Once again, a broad range of materials are available including Zn, Mg, Al, and Fe (Figure 3.3). More information can be found in several published reviews or guides. In the context of the ElectraSyn 2.0, micro electrodes are used for 1 or 2 mL vials and regular ones are used for other vial sizes. 2) Reference electrode When running a reaction with constant voltage, a reference electrode can be used to measure the true potential against a standard reference, which has a stable and well-­known potential. Running a constant potential reaction

Stain less steel

Various carbon electrodes

Figure 3.3  Electrodes.

Cu

Mg

Zn

Pt

Electrodes of different materials

Al

3.6 ­Electrasyn 2.

without this reference electrode only measures the non-­standardized potential between the cathode and anode [17]. 3) Purchasing electrodes Electrodes can be purchased from the IKA website. However, an electrode is simply an appropriately sized piece of pure metal that can be connected to an electrical source. Budget-­friendly electrodes can be obtained by purchasing metal sheets from Amazon and cutting them to the desired size for electrochemical flow or other applications. We tend to purchase RVC and carbon felt from outside suppliers such as Ultramet.

3.6.2  Interface 3.6.2.1  Hardware

On the ElectraSyn 2.0, there are 6 “touch” buttons and one knob (Figure  3.4). From top left to top right on the first line is the power touch button (1), the cyclic voltammetry button (2), and the home button (4). On the second row, there is the button to lock the screen (3), a button to view a graph of potential over the course of a constant current reaction or current over the course of a constant potential reaction (6), and the return button (5). The knob is used for scrolling within various menus and can be pressed to validate a choice.

(1) Power (2) CV (3) Lock

(4) Home (5) Return (6) Graph

Figure 3.4  ElectraSyn 2.0.

83

84

3  How to Confidently Become an Electrosynthetic Practitioner

3.6.2.2  Menus

1) Main menu Once the ElectraSyn 2.0 has been turned on via the power button, the main menu appears. It consists of three choices: ●● “Experiments” allows the user to choose from a previous experiment that has already been saved to the device by the user. ●● “New Experiments” allows the user to design a new experiment and leads them through the selection of various parameters. ●● “Assist Mode” can be used to help a user determine which current is best for their reaction. The general options “high,” “medium,” and “low” are listed, prompting the user to answer whether their reaction involves starting materials with a high, medium, or low reduction or oxidation potential. ●● On the top-­right corner of the screen is a gear symbol, which is the settings menu. 2) Settings Various settings can be modified within this menu. In “Communication,” you can change the language of the instrument. Perhaps the most important component of the settings is the voltage limit. Immediately upon receiving a new ElectraSyn, the user should change the voltage limit from 10 V to the maximum 30 V because the user will likely run a reaction (or set of reactions if using the carousel) that requires more than 10 V (Figure 3.5).

3.6.3  How to Set Up the Cell 1) Add a stir bar to the vial. If the reaction is air-­sensitive, wrap the threading of the vial with Teflon tape or similar to create a strong seal. Place the desired electrodes into the cap, making sure the electrodes are placed such that they are as close together as possible (Figure 3.6). When looking at the cap from the front, the anode should be on the left and the cathode should be on the right

Home page

Figure 3.5  Adjusting voltage limit.

Parameters

Voltage limit

3.6 ­Electrasyn 2.

Cathode

Anode

Standard electrode position

With reference electrode

Reaction under argon

Figure 3.6  Reaction setup.

when using a positive current. If desired, the reference electrode can be screwed into the middle (Figure 3.6). 2) As with any reaction, it is best to add solids first. Then, if working under inert atmosphere, you can place the vial under said atmosphere and add liquids through the septum. 3) Place the vial into the ElectraSyn square carrier or carousel. For various reasons, if the connection is poor, electricity will not flow properly. When placed correctly, the bottom of the vial should touch the black plate and you should not be able to twist the vial left to right, only slide it up and down (Figure 3.6). With older caps, be sure to check that there is not corrosion where the cap meets the ElectraSyn. Lastly, make sure that the square carrier or carousel is rotated all the way to the right. If the equipment is not rotated all the way to the right until it stops, the reaction will not work due to a poor connection (this is probably the culprit if the reaction is started, and the voltage maximizes immediately to 30 V). 4) If the contents are air-­sensitive, or a special reaction gas is desired, place a balloon filled with nitrogen, argon, or that gas on the vial (Figure 3.6).

3.6.4  How to Start an Experiment Select “New Experiment” on the main menu (Figure 3.7). From here, the instrument will prompt you through a few questions. ●●

Constant current or constant potential. In general, it is best to choose constant current. This is because as the reaction goes on, redox active species are consumed, resulting in an increase in the

85

86

3  How to Confidently Become an Electrosynthetic Practitioner

●●

resistance of the reaction. If a reaction is run at constant potential, the current will decrease as resistance increases such that the reaction becomes slower, less efficient, and lower yielding. One would consider using constant potential if there are sensitive functional groups that could be harmed by a reaction that has increasing potential as the resistance increases. Time, total charge, or run continuous. You can run the reaction for a set amount of time, until a set number of electrons are added, or until you as the user tell the instrument to stop. We recommend choosing the second option under almost all circumstances and will continue through this section as though the “total charge” option was selected. –– Running the reaction based on the total number of electrons is the most logical because the practitioner can determine how much current to apply based on the redox event happening in the vial. They can also decide how much excess to apply in this manner based on reaction efficiency. This parameter can be investigated during the optimization of the reaction.

(a)

1. Check the vial connection

2. Choose the experiment mode

3. Set the voltage or the current

4. Decide if you want to use a reference electrode

Figure 3.7  (a) Starting an electrochemical experiment. (b) Starting an electrochemical experiment.

3.6 ­Electrasyn 2.

(b)

5. Choose the time of the reaction

6. In total charge, put mmol of substrate and the number of faraday

7. Choose if you want to alternate the polarity

8. Decide if you want to save your experiment

9. Start the experiment

Figure 3.7  (Continued)

87

88

3  How to Confidently Become an Electrosynthetic Practitioner ●●

●●

●●

●● ●●

●●

Reaction scale and equivalents of electrons in Faradays per mol (F·mol−1). –– Once you enter the reaction scale in mmols, you must select the equivalents of electrons. Thinking about desired changes in oxidation state of the redox-­ active substances in solution can allow you to easily choose this value. For example, if a single-­electron reduction of the substrate should lead to the desired reaction, you can choose 1 F·mol−1. If there might be any inefficiency, you can apply more electrons that are needed, say 1.5 F·mol−1. For reactions in which a catalytic species will be affected by reduction or oxidation, you will want to adjust the F·mol−1 according to the change in oxidation state per substrate equivalent. Reference electrode. –– A reference electrode should be used if the reaction is being run at constant potential. The reference electrode measures the potential compared to a known standard. The IKA reference electrode is silver and can be used with KCl as a Ag/AgCl redox pair. If a constant potential reaction is run without a reference electrode, the instrument will measure and apply a potential based on the difference between the two electrodes in use. Alternating polarity. –– Alternating polarity can only be used when the cathode and anode are made of the same material. When this mode is selected, the direction of the current changes at the frequency indicated by the user. This can be used to prevent buildup at an electrode and has also been used as a tool to enhance selectivity and reactivity. Save experiment. Summary. If you put something in wrong, you can use the turning nob to adjust or check that parameter at this point. Start. –– Stir rate can be adjusted once the reaction is running. The ElectraSyn will automatically run at 400 rpm but increasing the stir rate to at least 1000 rpm is recommended because most electrochemical reactions are mass-­transfer limited.

3.6.5  During the Reaction 1) If the connection between the square carrier and ElectraSyn is faulty, the voltage will maximize (the other reason for this could be that the resistance of the solution is too high due to not enough electrolyte or a nonconducting solvent like THF). With the carousel, if the connection to the ElectraSyn is not correct, the electrolysis will not start. If there is a vial connection issue you will observe “–”on the screen at the position of the problematic vial.

3.6 ­Electrasyn 2.

2) Check the value of the voltage/­ current (Figure  3.8). From our experience, for a constant current experiment run at 5–40 mA, the resulting voltage should most likely be lower than 20 V. For a constant voltage experiment run at 2–20 V, the current should be below 40 mA if the reaction is properly set up. When running multiple vials at once on the carousel, the total voltage of the system can reach a maximum of 30 V. This means that to accommodate many vials, the system will automatically lower the applied current of the reaction if the ­corresponding total voltage of all Figure 3.8  During the reaction. the vials on the carousel is greater than 30 V. If you are concerned about high resistance in a reaction due to solvents that have a low dielectric constant or low concentration of electrolyte, do not use the carousel or do not fill the entire carousel. You can check for this by removing one of the vials and seeing if the voltage of the other vials increases. It is possible that the potential increases drastically over the course of the ­reaction. This is likely due to solids depositing on one or both electrodes. You can verify this after the reaction, but it indicates a problem with the reaction conditions such as solubility of products or salts issued from the sacrificial anode, plating-­out of metals onto the electrode, or unwanted side reactions. 3) Check electrodes for gas formation and plated metals or accumulation of organic materials. 4) Note changes of color. 5) Note solid precipitation. 6) Take aliquots of the reaction as desired for analysis by NMR, GC-­MS, TLC, etc.

3.6.6  After the Reaction 1) Check that the reaction was completed (all F·mol−1 have passed, or time of reaction is correct). 2) Note electrode state and reaction mixture state. 3) Clean vial like you would with any glassware. Remove the electrodes from the vial cap. Clean the cap with acetone and water if needed. Replace the septum if it is dirty. If connections appear rusted where the cap connects to the

89

90

3  How to Confidently Become an Electrosynthetic Practitioner

ElectraSyn or where the electrodes connect to the cap, wet a paper towel with 1 M HCl and scrub these areas. Electrodes can be cleaned with water, 1 M HCl, and acetone. Graphite and metal-­based electrode can also be cleaned with sandpaper. For more detailed descriptions of cleaning each type of electrode, IKA has a useful guide.1 In general, if an electrode looks dirty or darkened and cannot be cleaned, you should use a different one. Depending on the reaction being run, electrodes can expire quickly or be reused many times. As a chemist, it is best to use visual inspection to make this determination. If a reaction is suffering from reproducibility issues, it may also be time to use a fresh electrode. Porous materials like graphite and RVC tend to have shorter lifetimes while metallic electrodes tend to have longer ones. 4) Reaction analysis can proceed as desired.

3.7 ­Case Study Because electrochemistry requires electrodes and electrolyte as well as specific types of solvents, it can be daunting to translate a traditional chemical reaction into an electrochemical one. This section is meant to explain the logic behind the parameters of an electrochemical experiment, including reaction design and the optimization process. It describes a recent project developed in our group illustrating how a known photochemical reaction was achieved using electrochemistry. Through a combination of chemical intuition, knowledge of relevant precedents, and trial and error, we hope this case study enables the practitioner to confidently learn how to design an electrochemical experiment.

3.7.1  Project Overview Dialkyl ketones are a motif often found in biologically active molecules (Figure  3.9)  [19]. One way to synthesize dialkyl ketones is through reductive cross-­coupling (RCC). However, traditional RCC methods require the use of stoichiometric reductants [20–27]. Electrochemical RCC (ERCC) overcomes the use of chemical reductants in the synthesis of dialkyl ketones. Examples of ERCC for the synthesis of dialkyl ketones have been reported but are limited to the formation of benzylic or methyl dialkyl ketones and couple highly activated acylative reagents like acyl chlorides and anhydrides [28–30]. Imides are an alternative to more common reactive acyl precursors, and because of their stability, imides are especially useful as late-­stage acyl synthons in multistep syntheses. In the past, our lab has developed the coupling of imides with alkyl bromides using metallophotoredox catalysis to access ketones (Scheme  3.1a)  [31], but the need for an 1  https://www.ika.com/ika/pdf/flyer-­catalog/202103_Electrasyn%202.0_cleaning%20 electrodes_EN.pdf.

3.7 ­Case Stud

O

HO O

N

O

Inhibitor of heme oxygenase-1

H F

OAc

NH

O

Ph

H

O

Intermediate to bioactive scaffold

Ph O

O

H

N F

iPr

O

HO

Budesonide Anti-inflammatory O

O

OH

O OH

HO BzO AcO

O (R)-muscone Fragrance

Taxol® Chemotherapy medication

Figure 3.9  Occurrence of dialkyl ketones in bioactive molecules.

(a) O

O

O

Ni

Ir

+

N

R

Br

R Si

O 1°, 2° and 3° alkyl

3 examples up to 72% yield

Unactivated 2°

Use of expensive iridium photocatalyst

Use of high MW silane

Limited scope

(b) O R

X

O N

O 1°, 2° and 3° alkyl Undivided setup

O

Ni

+

Unactivated 1° and 2°

R

X = Br for alkyl and X = Cl for benzyl

Electrochemical conditions

30 examples

Access to diversified dialkyl ketones

Scheme 3.1  (a) Photochemical Ni-­catalyzed formation of dialkyl ketones from imides [31]. (b) Electrochemical nickel reductive cross electrophile coupling of imides with alkyl halides [32].

91

92

3  How to Confidently Become an Electrosynthetic Practitioner

expensive photocatalyst and stoichiometric silane reductant diminished the reaction’s utility. In 2022, through electrochemistry, we overcame these limitations (Scheme 3.1b) [32]. Key considerations for this project are described below.

3.7.2  Optimization of Parameters 3.7.2.1  Designing an Electrochemical Experiment

To design an electrochemical experiment, one should find a published reaction that is related to the desired transformation. However, this is not as simple as it sounds. The following hypothetical example illustrates the nuance in choosing the best model reaction: Imagine that one wishes to design an electrochemical reductive biaryl cross-­ ­ coupling of aryl tosylates with aryl chlorides. One could envision modeling the first experiment based on a hypothetical electrochemical ­ ­reductive cross coupling of alkyl carboxylic acids and alkenyl bromides. On the other hand, one could model the first experiment based on a hypothetical cross coupling between aryl tosylates and aryl chlorides with zinc as the ­terminal reductant. Although it might seem easier to transition one electrochemical reaction to another, we recommend using the reaction in which the desired transformation and corresponding substrates match most closely rather than one in which electrochemistry is used. Thus, the starting point for the electrochemical experiment design would be the latter, though it could also be beneficial to try the conditions of the former reaction with the desired substrates or to combine them as shown below. As a starting point for the work described in Scheme 3.1b, we selected the alkyl bromide and imide sources from our previous photochemical dialkyl ketone method  [32] and used the nickel source, solvent, and electrodes from a known electroreductive cross electrophile coupling reaction (Scheme 3.2) [33]. 1) Substrates Cyclohexane carbonyl succinimide was chosen as the acylative partner and 4-­bromotetrahydropyran was chosen as the alkyl source. Their ability to couple with nickel, and the desired equivalents of each coupling partner, was shown in a previous study [31]. An excess of alkyl halide is often required as it can undergo side reactions like homocoupling and protodehalogenation [34, 35]. 2) Catalyst A pre-­formed nickel(II) bromide 2,2′-­bipyridine complex was chosen as the nickel precursor because it is known to avoid plating of nickel at the cathode [36].

3.7 ­Case Stud

(a) +

R1 I

R2 Cl

Ni(bpy)2Br2, NBu4BF4, DMF, rt

R 1 R2

250 mA, Zn or Al(+)/Nickel foam(–)

(b) O

Br

O +

N

R

[Ir(dF(CF3)ppy)2dtbbpy]PF6 (1 mol%) [Ni(dtbbpy)(H2O)4)]Cl2 (3 mol%) Na2CO3 (1 equiv.), TMS3SiH (1 equiv.) Acetone or AcOEt, rt, 12–72 h, Blue LEDs

O

O R

(c) O

O N

O 3 (1 equiv.)

Br + O 4 (1.5 equiv.)

Ni(bpy)2Br2 (10 mol%) LiBr (2 equiv.) DMF (0.07 M), rt, Ar Zn(+)/Ni foam(–) 8 mA, 2 F · mol–1

O O 5

Scheme 3.2  From literature precedents to first electrochemical attempt. (a) Pioneering electrochemical Ni-­catalyzed formation of dialkyl ketones from acyl chlorides [33]. (b) Photochemical Ni-­catalyzed formation of dialkyl ketones from imides [31]. (c) Electrochemical proof of concept.

3) Electrolyte An electrolyte is a substance that contains ions and can conduct electricity. When an electrolyte is placed in a solution, the ions in the solution become mobile and can carry an electrical current through the solution. In organic solvents, common examples of electrolytes include salts such as tetrabutylammonium tetrafluoroborate (TBABF4), potassium hexafluorophosphate (KPF6), and lithium bromide (LiBr). Regarding our study, as bromide is already present in the media from alkyl bromide and the nickel source, LiBr was chosen as the electrolyte. 4) Additives No additives were added in the reaction as we could draw a viable catalytic cycle without any. 5) Solvent ERCC often employ polar aprotic solvents like DMF or DMA. Previously reported conditions indicated that DMF could be a suitable solvent for this reaction [33].

93

94

3  How to Confidently Become an Electrosynthetic Practitioner

6) Concentration In the 5 mL ElectraSyn vial, a volume of 3–4 mL is recommended to achieve proper surface contact between the solution and the electrodes. We first opted for a concentration of 0.1 M (which is common in cross coupling reactions) and a scale of 0.25 mmol of substrates to be engaged in the reaction (a scale where all analysis can be performed easily including any potential isolation steps), which corresponds to 2.5 mL of solvent. To have good surface contact between electrodes and the media, an extra milliliter of solvent was added, so the concentration was 0.07 M with 0.25 mmol of substrate and 3.5 mL of solvent. 7) Electrodes The desired cross coupling is purely reductive, meaning that only a reduction is required to turn over the catalytic cycle. Such reaction needs a setup with a cathode and a sacrificial anode (for more details regarding the different modes of electrochemical reactions, the reader is directed to Refs. [16–18]). Following the example of ERCC reactions in electrosynthesis from literature, a nickel foam cathode and a zinc sacrificial anode were chosen [32]. 8) Voltage and current As it is common when starting a new electrochemical reaction (please see Section 3.6.4), the constant current mode of electrolysis was adopted. For a scale of 0.25 mmol, we decided to apply a current value of 8 mA as most of the examples we found in the literature were applying currents between 5 and 10 mA. 9) Number of electrons Theoretically, two electron equivalents should be enough for our reaction to be completed, so 2 F·mol−1 were applied. At an early stage, we noticed that the starting material was not totally consumed after applying 2 F·mol−1, so we increased the F·mol−1 to 4. 10) Temperature Heating and cooling are difficult, though not impossible, to perform using ElectraSyn 2.0 without using the Gogo module accessory. To make screening faster, our reaction was run at room temperature.

3.7.3  Proof of Concept The final reaction conditions are shown in Scheme 3.3. The desired product was obtained in 36% yield determined by GC-­MS with an internal standard, and we proceeded with optimization from this point. 3.7.3.1  Optimization

While reaction parameters can be optimized in any order, depending on the ­reaction and what is known from literature precedent, there may be a logical order in which to conduct them. For our study, we first optimized the parameters

3.7 ­Case Stud

O

O N

O 3 (1 equiv.)

Br +

Ni(OAc)2 · 4H2O(10 mol%) bpy (10 mol%), KPF6 (2 equiv.) DMF (0.07 M), rt, Ar

O 4 (1.5 equiv.)

O

Zn(+)/Ni foam(–) 8 mA, 4 F · mol–1

O 5, 66% (57% isolated)

Scheme 3.3  Optimized conditions for the nickel-­catalyzed electroreductive cross coupling of imides with alkyl halides.

we expected to have the greatest impact on reaction yield. Thus, we started with electrolyte, followed by nickel catalyst source and ligand, the solvent, additives, equivalents of alkyl bromide, and finally electrochemical parameters including current, number of electrons, and electrodes. 1) Electrolyte Electrolytes are composed of a cation and an anion that can interfere with the catalyst. Therefore, we decided to first screen electrolytes. After screening a handful of classic salts, several parameters were considered to choose the most fitting one. Lithium perchlorate, tetrabutylammonium tetrafluoroborate, tetrabutylammonium hexafluorophosphate, and potassium hexafluorophosphate were all high yielding, but perchlorates have safety concerns. Tetrabutylammonium salts have a high molecular weight and are thus less atom-­economic. In addition, electrolytes such as tetrabutylammonium tetrafluoroborate and tetrabutylammonium hexaflurophosphate are expensive. Therefore, considering safety, molecular weight, and cost, potassium hexafluorophosphate (KPF6) was used for the rest of the optimization. Noteworthy, we decided to use two equivalents of electrolyte to insure sufficient conductivity that was assessed by looking at the voltage range obtained during the reaction (Table 3.1). 2) Precatalyst and ligand We screened pre-­ligated and non-­pre-­ligated nickel(II) bromide complexes and found bipyridine to be important (Tables 3.2 and 3.3). Then, other nickel sources were tested with this 2,2′-­bipyridine ligand. NiCl2•glyme and nickel(II) acetate tetrahydrate were both high yielding. Nickel(II) acetate tetrahydrate is 1/100th the price of NiCl2·glyme and NiBr2·glyme. Therefore, a pre-­ligated complex of nickel(II) acetate 2,2′-­bipyridine was attempted but afforded diminished yield. As we were interested in probing the mechanism during the precatalyst screening, bis(1,5-­cyclooctadiene)nickel, the pseudo-­ nickel(I) complex [(dtbbpy)NiCl]2, and a possible reaction intermediate were tested as potential active catalysts (entry 8–10). After this screening, the

95

96

3  How to Confidently Become an Electrosynthetic Practitioner

Table 3.1  Electrolyte screen. O

O

Br

N

+

O

Ni(bpy)2Br2 (10 mol%) Salt (2 equiv.) DMF (0.07 M), rt, Ar Zn(+)/Ni foam(–) 8 mA, 4 F · mol–1

O 4 (1.5 equiv.)

3 (1 equiv.)

O

O 5

Entry

Salt

Yield of 5a

1

LiClO4

60%

2

LiBr

34%

3

MgBr2

17%

4

TBAPF6

65%

5

TBABF4

63%

6

KPF6

62%

a

 Reactions were performed on a 0.25 mmol scale. Yields determined by GC-­MS using benzophenone as an internal standard.

Table 3.2  Pre-­catalyst screen. O

O N

+

O 3 (1 equiv.)

Nickel source (10 mol%) Ligand (10 mol%), KPF6 (2 equiv.) DMF (0.07 M), rt, Ar

O 4 (1.5 equiv.)

O

Zn(+)/Ni foam(–) 8 mA, 4 F · mol–1

O 5

Entry

Nickel source

Ligand

Yield of 5a

1

Ni(bpy)2Br2

—

62%

2

Ni(bpy)3Br2

—

55%

3

NiBr2· glyme

bpy

69%

4

NiBr2· glyme

dtbbpy

36%

5

Ni(OAc)2· 4H2O

bpy

66%

6

NiCl2· glyme

bpy

69%

7

Ni(dtbbpy)(OAc)2· 2H2O

—

50%

8

[Ni(dtbbpy)Cl]2

—

29%

9

Ni(cod)2 Ni(dtbbpy)(COCy)(succ)

bpy —

46% 42%

10 a

Br

 Reactions were performed on a 0.25 mmol scale. Yields determined by GC-­MS using benzophenone as an internal standard.

3.7 ­Case Stud

Table 3.3  Ligand screen. O

Ni(OAc)2 · 4H2O (10 mol%) Br Ligand (10 mol%), KPF6 (2 equiv.) DMF (0.07 M), rt, Ar

O N

+

O 3 (1 equiv.)

O

Zn(+)/Ni foam(–) 8 mA, 4 F · mol–1

O 4 (1.5 equiv.)

O 5

Ligand Ph

Ph

N

tBu N

N

54%

N

tBu

N

66%

N

38%

20 mol% ligand: 46 mol% 30 mol% ligand: 39 mol% Me

Me Me N

N

54%

N

N

46%

Me Me

N

N Me

14%

Reactions were performed on a 0.25 mmol scale. Yields determined by GC-­MS using benzophenone as an internal standard.

nature of the nickel was found to only have a minor impact on the reaction yield. Thus, nickel(II) acetate tetrahydrate was chosen as a cost-­effective nickel source. On the other hand, the ligand clearly impacted the yield of the reaction. Only nitrogen-­based ligands were tested as they are reported to be especially effective for nickel-­catalyzed acylative RCC [28–30]. The steric properties of the ligand were significant; 6,6′-­dimethyl-­2,2′-­dipyridyl only gave14% yield. Of the bipyridine ligands tested, 2,2′-­bipyridine gave the highest yield and is a cost-­effective option. Higher equivalents of ligand compared to nickel afforded diminished yield, so equimolar equivalents of 2,2′-­bipyridine and nickel were adopted. 3) Solvent Polar solvents were explored because resistance in solvents such as THF, hexanes, toluene, and DCM is too high to run the reaction at a reasonable potential due to low solubility of salts in those solvents as well as lower dielectric constant (Table 3.4). Out of all the screened solvents that afforded high yields

97

98

3  How to Confidently Become an Electrosynthetic Practitioner

Table 3.4  Solvent screen. O

O N

Br +

O 3 (1 equiv.)

Ni(OAc)2 · 4H2O (10 mol%) bpy(10 mol%), KPF6 (2 equiv.) Solvent (0.07 M), rt, Ar

O 4 (1.5 equiv.)

O

Zn(+)/Ni foam(–) 8 mA, 4 F · mol–1

O 5

Solvent

Yield of 5a

1

DMF

66%

2

DMSO

60%

3

DMAc

41%

4

AcCN

5%

Entry

a

 Reactions were performed on a 0.25 mmol scale. Yields determined by GC-­MS using benzophenone as an internal standard.

of the desired ketone product, DMF was the cheapest one and was thus preferred. Inspired by our previous work, concentration was chosen to be 0.1 M [31]. However, an extra milliliter was added to the initial 2.5 mL of solvent to obtain an optimal contact surface between electrodes and the media, leading to a reaction concentration of 0.07 M. 4) Additives Some previous reports involve additives to enhance the yield, and we tested these additives as well (Table 3.5) [37–40]. First an iodide source was tested to promote halogen exchange with the nickel catalyst, but this diminished the yield. Potassium carbonate, a coordinating base, was added, but it did not significantly impact the yield. Magnesium salt, a coordinating species, and lithium bromide, an additional source of bromide in the media, also afforded diminished yield. Catalytic amount of zinc salts, added at the start of the reaction, were proven to enhance catalyst activity and to increase the yield of RCC reactions [38]. Under our electrochemical reaction conditions, this was not the case. Water can also help to solubilize some ionic species. In our case, catalytic amounts of water were slightly detrimental, showing that the reaction is sensitive to water. Finally, an electron shuttle has been described in ERCC reactions to avoid overreduction of substrates by interacting with the cathode in place of the other reaction components [41]. The shuttle gets reduced at around –1.8 V vs Fc/Fc+ and can be added to avoid reduction below this potential  [41]; 20 mol% of Ni(MeBPI)2 shuttle was added to the reaction. Although it avoided nickel plating at the cathode, showing that it did act as an electron shuttle, no change in yield was observed, which demonstrates that overreduction is not a limiting parameter.

3.7 ­Case Stud

Table 3.5  Additives screen.

O

O N

Br +

O 3 (1 equiv.)

Ni(OAc)2 · 4H2O (10 mol%) bpy(10 mol%), KPF6 (2 equiv.) Additive (1 equiv.) DMF (0.07 M), rt, Ar

O

Zn(+)/Ni foam(–) 8 mA, 4 F · mol–1

O 4 (1.5 equiv.)

O 5

Entry

Addtive

Yield of 5a

1

None

66%

2

NaI

28%

3

K2CO3

57%

4

MgCl2

30%

5

LiBr

35%

6

ZnBr2 (10 mol%)

53%

7

H2O (20 mol%)

47%

8

Ni(MeBPI)2 (20 mol%)

61%

a

 Reactions were performed on a 0.25 mmol scale. Yields determined by GC-­MS using benzophenone as an internal standard.

5) Substrate equivalents The importance of the ratio between coupling partners has been highlighted in  recent studies (Table  3.6)  [42]. The optimal ratio was reached at 1  :  1.5 imide : alkyl bromide. Table 3.6  Equivalents screen. O

O N

+

O 3 (1 equiv.) Entry

a

Br

O 4 (X equiv.)

Ni(OAc)2 · 4H2O (10 mol%) bpy(10 mol%), KPF6 (2 equiv.) DMF (0.07 M), rt, Ar

O

Zn(+)/Ni foam(–) 8 mA, 4 F · mol–1

X equiv. of 4

O 5 Yield of 5a

1

1

43%

2

1.5

66%

3

2

48%

 Reactions were performed on a 0.25 mmol scale. Yields determined by GC-­MS using benzophenone as an internal standard.

99

100

3  How to Confidently Become an Electrosynthetic Practitioner

6) Electrochemical parameters Of electrochemical parameters, the current should be optimized first (Table 3.7). We found 8 mA to be the best. As expected, the absence of electricity did not lead to product formation. This control should always be performed to ensure that electricity is required to perform the desired reaction. The number of electron equivalents needed for the reaction to be high yielding was 4 F·mol−1 (Table 3.8). Fewer equivalents give a lower yield and greater equivalents did not improve the yield. Finally, different electrode pairs were tested (Table 3.9). First the zinc sacrificial anode was replaced by magnesium as such electrode was successfully used in the context of reductive cross coupling reactions [33]. Unfortunately, this sacrificial anode only provided trace amount of desired product. This is certainly due to the formation of undesired by-­products issued from the reaction between Mg cations and DMF as demonstrated by the group of Périchon in one of their electrochemical studies [43]. An attempt to replace the sacrificial anode by a combination of amine (added as a sacrificial reductant, see Ref. [44] for more details) with a non-­ sacrificial anode of graphite, only gave traces of product. Then, different cathodes were explored. Graphite gave very low yield, probably due to its low surface area. An RVC cathode gave good yields, but nickel foam was better. As such, nickel foam cathode and zinc anode provided the best results.

Table 3.7  Current variations. O

O N

+

O 3 (1 equiv.)

a

Br

Ni(OAc)2 · 4H2O (10 mol%) bpy(10 mol%), KPF6 (2 equiv.) DMF (0.07 M), rt, Ar

O 4 (1.5 equiv.)

O O

Zn(+)/Ni foam(–) Current, 4 F · mol–1

5

Entry

Current

Yield of 5a

1

4 mA

54%

2

6 mA

53%

3

8 mA

66%

4

10 mA

64%

5

12 mA

60%

6

No electricity

0%

 Reactions were performed on a 0.25 mmol scale. Yields determined by GC-­MS using benzophenone as an internal standard.

3.7 ­Case Stud

Table 3.8  F·mol−1screen. O

O N

Br +

O 3 (1 equiv.)

Ni(OAc)2 · 4H2O (10 mol%) bpy(10 mol%), KPF6 (2 equiv.) DMF (0.07 M), rt, Ar

O 4 (1.5 equiv.)

O O

Zn(+)/Ni foam(–) 8 mA, X F · mol–1

5

Entry

X F · mol–1

Yield of 5a

1

3

42%

2

4

66%

3

5

60%

a

 Reactions were performed on a 0.25 mmol scale. Yields determined by GC-­MS using benzophenone as an internal standard.

Table 3.9  Electrode screen. O

O N

Br +

O 3 (1 equiv.)

Ni(OAc)2 · 4H2O (10 mol%) bpy(10 mol%), KPF6 (2 equiv.) DMF (0.07 M), rt, Ar

O 4 (1.5 equiv.)

O

Anode(+)/Cathode(–) 8 mA, 4 F · mol–1

O 5

Entry

Anode(+)/Cathode(–)

Yield of 5a

1 2

Zn(+)/Ni foam(–) Mg(+)/Ni foam(–)

66% 0.0010>CCOC(=O)c1nc(Br)sc1C.CCc1[nH]c(C(= O)N[C@H]2CCN(C(=O)OC(C)(C)C)C[C@H]2OC)nc1C(F)(F)F|CCc1[nH]c(C(=O)N[C@H]2CCN(C(=O)OC(C)(C)C)C[C@H]2OC)nc1C(F )(F)F>0.9998>CO[C@@H]1CN(C(=O)OC(C)(C)C)CC[C@@H]1N.CCc1[nH]c(C(=O)O)nc1C(F)(F)F|CCc1[nH]c(C(=O)O)nc1C(F)(F)F>0. 9916>CCc1[nH]c(C=O)nc1C(F)(F)F.[O-][Cl+][O]|CCc1[nH]c(C=O)nc1C(F)(F)F>0.9994>CCc1[nH]c(C(OC)OC)nc1C(F)(F)F|CCc1[nH]c(C(OC)OC)nc1C(F)(F)F>1.0000>N.O=C1CCC(=O) N1Cl.COC(C=O)OC.CCC1(C(=O)C(F)(F)F)SCCCS1|CCC1(C(=O)C(F)(F)F)SCCCS1>1.0000>CCC1SCCCS1.CCOC(=O)C(F)(F)F', 'route_cost': 6.905859674323364, 'route_len': 6}

Figure 11.10  A sample printout by running the example.py script under Retro* repo. The routes are produced in alternating sequence of molecule sets and confidence scores associated with the single steps.

11.4.2  Closed-­Source Tools There are a growing number of commercial CASP tools that are not open-­sourced and typically require a paid license to use. Because the authors of this chapter do not have access to all of these software packages (and because license agreements for scientific software in general may preclude direct evaluations or comparisons), we are limited in the amount of detail we can include in this section. Additionally, each platform listed is a complex piece of software with a unique user interface and set of features that are difficult to capture in any quantitative manner.

443

444

11  Computer-­Assisted Synthesis Planning

Table 11.1  A list of closed-­source CASP programs under active commercial development. Literature references to the methods employed in each program are included where available. Tool

Company

Refs.

Demo availability

Synthia

Merck KGaA

[8, 26, 97]

Upon request

RXN

IBM

[64, 98]

Free

Predictive Retrosynthesis Planner

Reaxys

[7]

Free (limited capabilities)

SciFindern

CAS

ICSynth

DeepMatter

[99]

Upon request

Molecule.one

Molecule.one

[43, 100]

Upon request

Spaya

Iktos

Manifold

PostEra

Upon request

Free [51]

Free

We attempt to provide a brief summary of the different commercial CASP tools listed in Table  11.1 to provide readers with a sense of the diversity of available software tools. Where we can, we have attempted to highlight the datasets that each tool is developed from and links back to as well as the methods used to generate retrosynthetic plans. We do not discuss the details of each program’s interface as this would be best assessed by a user directly. Synthia (formerly known as Chematica) differentiates itself from other CASP tools by its database of expert-­curated reaction rules. An example reaction template from Synthia is shown in Figure 11.2 in Section 11.2.2, illustrating how the templates explicitly encode various aspects of reaction selectivity, incompatibility, and potential cross-­reactivity. Synthia now includes over 100 000 such retrosynthetic reaction templates. For multistep synthesis planning, Synthia applies these templates iteratively to search for precursors that maximize a user-­customizable scoring function. The Synthia multistep search algorithm allows the search to sometimes step to precursors that have a worse score if they allow the algorithm to reach better precursors further along the search [8]. Synthia includes a building block catalog from Sigma Aldrich and over 80 other vendors with over 4.2 billion compounds and their corresponding catalog price. Synthia has been shown to find promising pathways to complex organic molecules including pharmaceuticals and natural products, discussed in Section 11.5.3. The chemistry database Reaxys offers an integrated Predictive Retrosynthesis Planner. The planning tool can piece together known reactions from the Reaxys database, as well as generalize known reaction chemistry to novel products. The synthesis planner applies the “neural-­symbolic” [7] approach from Segler et al. to single-­step

11.4  ­Select Examples of Software Tools for CAS

retrosynthesis. This means that a neural network is trained to prioritize a set of ­retrosynthetic templates for an input product molecule. The templates are algorithmically extracted from the Reaxys reaction database, which of course will continue to grow. The planner incorporates additional information for reaction steps from the Reaxys database such as yield (when known), and reaction steps in returned routes link directly to entries in Reaxys. The buyable stopping points for the search are defined from multiple vendor catalogs and can be customized by the user. CAS similarly offers Scifindern, a retrosynthesis tool integrated with the CAS content collection. Scifindern also enables users to include a mix of exact known reaction steps as well as known templates applied to new products. Evidence is provided for each proposed reaction step from the extensive CAS reaction database. Multistep retrosynthesis searches are stopped once buyable starting materials are reached with a user-­defined price limit. In addition to common search parameters like search depth, SciFindern allows users to specify which bonds to break or to maintain in the retrosynthetic search. The tool also allows users to select whether to include reaction types with few precedents when considering predicted retrosynthetic steps or whether to consider only well-­supported chemical transformations. IBM’s RXN is a free-­to-­use but closed-­source CASP platform that uses template-­ free Transformer-­based retrosynthesis models. These models – adapted from the field of natural language processing – are trained to predict precursor molecules directly from an input product molecular structure, with no intermediate prediction of a reaction template. RXN’s chemical reaction transformer models are trained on a dataset that was generated from patents in the United States Patent and Trademark Office (USPTO) as well as the Pistachio database from NextMove Software, which is also derived from the patent literature but includes European patents. As a translation-­based tool, IBM RXN is particularly sensitive to the data source on which it is trained, as there is no chemical intuition built into the models. IBM recently announced a partnership with Science of Synthesis/Thieme Chemistry, who curate and maintain databases of chemical reactions. ICSynth, acquired by DeepMatter, uses a set of retrosynthetic “transforms”; templates algorithmically extracted from reaction databases that include InfoChem’s own SPRESI, Thieme’s Science of Synthesis, and the enzymatic reaction database Rhea. Additionally, users can input custom sets of reactions from which transforms are automatically generated. ICSynth links to the eMolecules and Chemspace databases of buyable building blocks. The reported algorithm for ICSynth [99] first assesses which bond transformations from the target molecule would lead to a simpler precursor and then searches for retrosynthetic transforms to achieve those bond changes, although details are sparse. Many new commercial CASP tools have been developed in recent years, leveraging new machine learning models, search algorithms, or prioritization techniques

445

446

11  Computer-­Assisted Synthesis Planning

for discovered synthesis plans. Beyond these handful of examples are several more commercial programs, including molecule.one (graph-­based), Iktos’s Spaya (template-­based), PostEra (translation-­based), and Galixir’s Pyxir (founders affiliated with GLN and Retro*). Internal efforts are underway, anecdotally, at dozens of companies who are building their own internal pipelines for synthesis planning even if not selling software directly. We mention these tools briefly but do not mean to understate their utility and importance to the field; chemists interested in finding the CASP system that works best for them should try out a number of programs and find which workflows are most comfortable and suitable for their use case. Ultimately, the best way to assess these different tools is to directly test out their demo versions to see which tool offers the right capabilities for one’s project goals. Some considerations we believe to be important are summarized in Section 11.4.4.

11.4.3  CASP Tools for Enzymatic Catalysis While CASP has a long history for conventional organic synthesis, its application to “retrobiosynthesis” [101] is more recent. However, as manipulating enzymatic chemistry in cells and in vitro has become easier (see Chapter 3), interest in tools to automate the design of novel metabolic pathways and biocatalytic cascades has surged. The underlying algorithms and model architectures are similar to organic chemistry CASP tools, but retrobiosynthesis CASP tools are trained on datasets of enzyme-­catalyzed reactions, often use metabolite databases instead of vendor catalogs to define building blocks, and can include additional considerations such as estimated catalytic promiscuity for a given enzyme sequence. Examples of user-­accessible retrobiosynthesis tools include RetroPathRL  [102], BNICE. ch [103], and RetroBioCat [104]. RetroPathRL and BNICE.ch are tuned toward metabolic engineering projects in living cells; they employ reaction templates extracted from MetaNetX and the Kyoto Encyclopedia of Genes and Genomes (KEGG), respectively. Alternatively, RetroBioCat is tuned toward planning of industrial biocatalytic cascades; the platform uses a set of manually encoded reaction templates that correspond to well-­explored biocatalytic transformations. In addition to these dedicated retrobiosynthetic tools, organic chemistry CASP tools like RXN, ICSynth and ASKCOS have also incorporated reaction data from various enzymatic reaction datasets [105].

11.4.4  Practical Considerations for CASP Programs We believe that having a basic understanding of how CASP tools work under the hood helps navigate their settings and proposals more effectively. For example, understanding that a data-­driven template-­based program may fail to generate an expected idea could be a result of missing the necessary template, having that

11.4  ­Select Examples of Software Tools for CAS

template be ranked poorly, or having an insufficiently exhaustive tree search that cannot find the right starting materials, which may or may not be in the buyable database. If no results are found during a multistep expansion, examining one-­ step recommendations using more permissive settings can help diagnose the root cause. There are a few additional factors we feel warrant more discussion: whether one can connect suggestions to literature precedent, whether one can access the program through a graphical interface, whether using a program requires transmitting structural information to the software provider, and whether one can customize the tool to fit the specific needs of a user or organization. 11.4.4.1  Traceability to Literature Precedent

In our experience, an important feature of a retrosynthetic planning tool for a chemist is the ability to check a suggested reaction or retrosynthetic step against precedents from a database and, ultimately, from published literature. Different tools vary in which databases they link to and how they link to them. For example, in the template-­based retrosynthesis tools, such as Reaxys Predictive Retrosynthesis or ASKCOS, each suggested retrosynthetic step corresponds to a retrosynthetic reaction template, which is linked to a set of precedent reactions. On the other hand, for tools that rely on template-­free methods, such as IBM’s RXN, there is no direct link between a suggested retrosynthetic step and database precedents. Instead, in RXN, precedents from the USPTO or Pistachio database are linked to a suggestion based on the structural similarity of the database example and the proposed reaction; for most use cases, simply searching a reaction database for entries similar to the proposed reaction may be a sufficient way to cross-­reference the literature. 11.4.4.2  How to Use CASP: Command Line Versus Graphical User Interface

Synthetic chemists may or may not be comfortable with programming concepts, package installation/management, and the use of the command line to run these programs without an interface. Most open-­source tools use Python, fortunately, so the barrier to entry is perhaps lower than it might otherwise be for other languages. Regardless, some free tools and all/most commercial tools are accessed through a GUI and therefore require no knowledge of coding whatsoever. GUIs help users navigate the vast number of ideas CASP tools may generate and make it easier to visualize and sift through pathway options. For chemists who wish to plan routes to only one or a few molecules, this is an ideal workflow. For other use cases in medicinal chemistry, one may wish to examine thousands of potential molecules and triage them based on perceived synthesizability; in these cases, programmatic access through the command line or an application programming

447

448

11  Computer-­Assisted Synthesis Planning

interface (API) is necessary, i.e. writing code. It should not be surprising that commercial CASP tools invest a great deal into their interfaces, as the user experience is an important aspect of gaining traction and adoption. For users who are interested in testing a wide variety of approaches including newly published methods from academia, familiarity with Python and the command line will be necessary. 11.4.4.3  Data Privacy

Depending on one’s use case, maintaining the confidentiality of target molecules might be essential. If a CASP tool is only available through a web application, or if the computation occurs on the software provider’s servers, information security should be a concern unless one does not intend to file intellectual property claims on that molecule. Certain programs, including but not necessarily limited to open-­source ones, can be deployed behind an organization’s firewall without any communication with external servers. This introduces the need to provide the necessary computational resources oneself, but that tradeoff may be well worth it to most pharmaceutical companies. 11.4.4.4  Customization Ability

Related to the notion of data privacy is the question of whether one can retrain or adapt CASP model(s) to one’s own data, with or without sharing that information with a third party. Data-­driven CASP models can in principle be retrained or fine-­ tuned on new reaction datasets to tailor their predictions to the distribution of reaction types a user is most interested in; whether there are convenient scripts and extract-­load-­transform pipelines to facilitate retraining is a separate matter. The most common customization users desire, in our experience, is the definition of a “solved” molecule during retrosynthetic analysis. Programs for multistep planning will be shipped with some default database of buyable (or otherwise appropriate) starting materials. Users may wish to supplement this database with the internal chemical inventory at their organization, or with a larger collection corresponding to a more inclusive price threshold. If customization would benefit one’s use case, we recommend evaluating candidate CASP tools along this dimension. Open-­source tools provide the greatest flexibility, at least in principle.

11.5 ­Case Studies As mentioned in the Introduction, retrosynthetic planning tools have reached an inflection point where they are starting to be used with greater success and greater popularity than ever before. Three case studies are given in the following sections to highlight the rapid progression of this field in recent years.

11.5 ­Case Studie

11.5.1  Segler et al.’s Data-­driven Program and A/B Testing Success In 2018, Segler, Preuss, and Waller published their landmark paper describing the successful development of a data-­driven retrosynthetic planner with the most compelling performance evaluation seen up to that point [7]. This program used a Monte Carlo tree search (MCTS) strategy to balance exploration and exploitation, following a neural network policy trained on templates algorithmically extracted from the Reaxys database. Among their evaluations was a medium-­scale analysis of roughly 500 products reported in the literature, which showed that the MCTS algorithm found hypothetical routes to a greater proportion (92%) of those compounds given a 60-­second computational budget as compared to a neural best-­first search (71%) or heuristic best-­f irst search (4%). Perhaps the more impressive evaluation, however, was a double-­blind preference test conducted with 45 graduate-­level organic chemists. Nine synthetic pathways to nine products were extracted from the literature and presented alongside nine model-­predicted pathways to the same targets. Each chemist participant was asked to select the preferred route based on their qualitative understanding of feasibility and personal preference. The authors found there was a slight, albeit not significant, preference toward the program’s generated route compared to the literature route on average. This was the first notable attempt at demonstrating the quality of routes proposed by a data-­driven retrosynthetic planner, taking advantage of machine learning advances from just a few years prior. While the software reported in the paper was never open sourced or made available, it has since been adapted and integrated into the commercial Reaxys Predictive Retrosynthesis model offered by Elsevier.

11.5.2  MIT’s ASKCOS Program and Robotic Synthesis Demonstration A similar template-­based retrosynthetic planner features in the ASKCOS software suite, which encompasses the additional CASP tasks of condition recommendation and reaction outcome prediction. While components had been published in earlier work, ASKCOS was first formally disclosed in a 2019 publication describing its integration with laboratory automation  [66]. This paper illustrated an approach to target-­oriented synthesis wherein synthetic routes could be planned by a data-­driven program and executed on a robotic flow chemistry platform. While there are several limitations to be aware of, the integration of AI planning and robotic synthesis brought together two research threads that had to-­date been advancing in parallel. A total of 15  medicinally relevant small molecules were

449

450

11  Computer-­Assisted Synthesis Planning

synthesized in continuous flow, each requiring 1–3 synthetic steps, including two molecules with defined stereochemistry. The limitations of this demonstration connect back to the challenge of having incomplete data on chemical reactions with which to train predictive models. In Segler et  al.’s evaluation, routes were examined by chemists on paper and only included the identities of reactants and reagents that contribute heavy atoms to  the product. For experimental synthesis, one additionally requires all of the additional procedural details to carry out the synthesis. As mentioned earlier, these considerations include far more than what can be predicted by the current generation of condition recommendation tools, which lack the ability to predict quantitative concentration information, reaction times, or purification steps and lack an understanding of flow compatibility, among other factors. Nevertheless, data-­driven tools like ASKCOS will almost certainly transition toward a more quantitative mode of operation as data availability improves. This transition would further facilitate the ability to accelerate target-­oriented chemical synthesis. For now, the open-­source ASKCOS tool provides a starting point for further route development by expert chemists.

11.5.3  Grzybowski’s Chematica/Synthia Program’s Experimental Validations and Acquisition Grzybowski and coworkers have made a number of invaluable contributions to the field of computer-­aided synthesis planning over the past decade. They have championed the expert approach to retrosynthesis wherein human chemists curate detailed transformation rules (and exceptions) that cover as many known reaction types as possible. Their set of expert templates form the basis of the retrosynthetic engine within the Chematica program, which is summarized and contextualized in a review by Szymkuc et al. [26]. A major milestone in its validation came in 2018 when a team of chemists from Grzybowski’s team and MilliporeSigma reported a joint study where Chematica-­planned routes to eight challenging targets were validated experimentally [97]. These targets were considered challenging as each had been synthesized either in low yield, with poor scalability, or entirely unsuccessfully. The experimental execution of these routes required a reasonable amount of condition screening and optimization, which is to be expected given that only coarse template-­level conditions are suggested, rather than substrate-­specific conditions. The success of this demonstration led to the purchase and rebranding of Chematica as Synthia by MilliporeSigma, which is now a commercial offering. Further validation of Synthia was reported using a suite of complex natural products, rather than smaller medchem-­like compounds  [8]. The reaction types  used for natural product synthesis do not necessarily resemble the

11.6 ­Conclusio

transformations that appear most often in reaction databases, so these structures are quite challenging for data-­driven techniques. Expert-­encoded templates have the potential to generalize more effectively as humans can describe directly when/if the rule should be applied. Of the many total syntheses proposed by Synthia, three were experimentally validated, though again requiring condition screening and optimization. A preference test was performed in a manner almost identical to Segler et al. to gauge whether program-­proposed routes were identifiable as more likely to be human-­designed or program-­designed when displayed side by side by literature (validated) pathways. The result was likewise the same: experts could not tell when a pathway came from the program or the literature. This represents a major accomplishment for Synthia and the field, as these natural product targets were rather complex; the full test set and list of pathways are available in their supporting information. An additional case study involving the use of Synthia for route planning and validation was reported by Lin et al., who applied the retrosynthetic tool to find routes to several antiviral compounds [106]; this paper provides a useful discussion of the successes and failures of computer assistance from a practitioner’s perspective.

11.6 ­Conclusion The wealth of options when considering using CASP tools to aid in target-­oriented synthesis (particularly, for retrosynthetic analysis) is a direct reflection of the field’s rapid growth and maturation. While these tools have existed in some form for decades, we do believe that the field hit an inflection point around 2017–2018 that coincided with the demonstration and/or experimental validation of both data-­driven and expert-­driven methods (Figure 11.11). There are many ways in which data science and machine learning will continue to influence research in the chemical sciences, and CASP is an area ripe for further growth. Even if a chemist is not interested in paying for a commercial system, free-­to-­ use platforms are highly competitive in the data-­driven category. What open-­ source tools might lack in certain usability features (or customer support) that can improve the user experience, they might make up for in ease of customization and internalization into molecular design workflows. There are plenty of new methods for retrosynthesis emerging seemingly every few months at least, although many are not integrated into a larger multistep path planning program with an interface. These tools are challenging or impossible to compare in terms of raw “performance” due to the complexity of evaluating retrosynthetic pathways quantitatively, so we have instead focused on the relationship between their approaches and the few examples of validation that prove their collective utility.

451

452

11  Computer-­Assisted Synthesis Planning

(a)

(b)

O O N

OH O

H N

OH

O

O O O

O

S

Safinamide

O O

O

F

O

N

F

F N

N

N H

NH

O

NH2

N

F

O

O

H N

N

O

N

F

N O

N

Celecoxib

N

(c)

OH

O

F

N

S

O HN

S

N

NH F

Quinapril

(d)

S

O

N

O

O N

N

N

O

O

O

O HO

S

N

N

HO

N

NH

O

O

Lidocaine

O

NH2

N H

F

O

HO

N

N

O

O

Cl

N

O

O

LP99 OH O

Dauricine

O

O

β-hydroxylurasidone S

HO

O O OH

N H

N

OH H

OH

N O O

OH

Engelheptanoxide C

(S)-4-hydroxyduloxetine

Lamellodysidine A

(R,R,S)-tacamonidine

Figure 11.11  Examples of molecular targets from the case studies described in Section 11.5. (a) Targets from [7] whose predicted synthesis plan was rated higher than the corresponding literature synthesis route. Source: Segler et al. [7]/Springer Nature. (b) Targets from [66] whose synthesis plan was automatically generated with ASKCOS and executed successfully by a robotic synthesis platform. Source: Coley [66]/American Association for the Advancement of Science. (c) Drug molecules from [97] and (d) natural product molecules from [8] whose synthesis plans were automatically generated with Chematica and successfully executed by chemists. Sources: (c) Klucznik et al. [97]/ Elsevier; (d) Mikulak-­Klucznik et al. [8]/Springer Nature.

Admittedly, much is missing from the current generation of CASP programs. Predicting the stereoselectivity of reactions can be difficult, as can be predicting the full substrate tolerance of a synthetic transformation with little experimental precedent. Data-­driven methods work best on compounds that resemble compounds they have been trained on, which are biased toward flat, sp2-­rich med chem-­like compounds; programs trained on data from journal articles may have a greater diversity of reactions in their knowledge base than programs trained only on patent information, although medicinal chemistry use cases are well addressed by both. Complex natural products are better left to expert programs like Synthia,

  ­Reference

at least for now, as the models behind data-­driven methods have yet to be fine-­ tuned for this use case. Beyond retrosynthesis, there is also plenty of scope for future work in condition recommendation and reaction outcome prediction. These models are likewise already at a useful state but will need to transition to be more quantitative than qualitative in order to drive adoption. Condition recommendation in particular can provide good initial guesses for what species to use, but does not address the challenge of finding appropriate concentrations or equivalence ratios. Reaction outcome prediction does not inform yields (except for specific machine learning models trained to predict yields from narrow high-­throughput experimentation campaigns), although it can predict relative selectivities in certain cases. Here, we have focused on models with the broadest possible domain of applicability (i.e. “any” organic transformation), but there are many examples of predictive models built on smaller, more focused datasets that represent a complementary part of the data chemist’s toolbox. Different types of models will prove useful in different settings, so we hope this chapter has left the reader with a sufficient understanding and interest in these techniques to at least give them a try.

Key References Research articles cited in Section  11.5 Case Studies will provide the interested reader with a strong understanding of the most significant themes in the field.

­References 1 Davies, I.W. (2019). The digitization of organic synthesis. Nature 570 (7760): 175. Ott, M.A. and Noordik, J.H. (1992). Computer tools for reaction retrieval and 2 synthesis planning in organic chemistry. A brief review of their history, methods, and programs. Recl. Trav. Chissm. Pays-­Bas 111 (6): 239–246. 3 Todd, M.H. (2005). Computer-­aided organic synthesis. Chem. Soc. Rev. 34 (3): 247–266. 4 Cook, A., Johnson, A.P., Law, J. et al. (2012). Computer-­aided synthesis design: 40 years on. Wiley Interdiscip. Rev. Comput. Mol. Sci. 2 (1): 79–107. 5 Ravitz, O. (2013). Data-­driven computer aided synthesis design. Drug Discov. Today Technol. 10 (3): e443–e449. 6 Warr, W.A. (2014). A short review of chemical reaction database systems, computer-­aided synthesis design, reaction prediction and synthetic feasibility. Mol. Inform. 33 (6–7): 469–476.

453

454

11  Computer-­Assisted Synthesis Planning

  7 Segler, M.H.S., Preuss, M., and Waller, M.P. (2018). Planning chemical syntheses with deep neural networks and symbolic AI. Nature 555 (7698): 604–610.   8 Mikulak-­Klucznik, B., Gołębiowska, P., Bayly, A.A. et al. (2020). Computational planning of the synthesis of complex natural products. Nature 588: 83–88.   9 Engkvist, O., Norrby, P.-­O., Selmi, N. et al. (2018). Computational prediction of chemical reactions: current status and outlook. Drug Discov. Today 23 (6): 1203–1218. 10 Johansson, S., Thakkar, A., Kogej, T. et al. (2020). AI-­assisted synthesis prediction. Drug Discov. Today Technol 32–33: 65–72. 11 Gao, W., Raghavan, P., and Coley, C.W. (2022). Autonomous platforms for data-­driven organic synthesis. Nat. Commun. 13 (1): 1075. 12 Schneider, G. (2018). Automating drug discovery. Nat. Rev. Drug Discov. 17 (2): 97–113. 13 Weininger, D. (1988). SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules. J. Chem. Inf. Comput. Sci. 28 (1): 31–36. 14 Rogers, D. and Hahn, M. (2010). Extended-­connectivity fingerprints. J. Chem. Inf. Model. 50 (5): 742–754. 15 Gilmer, J., Schoenholz, S.S., Riley, P.F. et al. (2017). Neural message passing for Quantum chemistry. Proc. 34th Int. Conf. Mach. Learn 70: 1263–1272. 16 Veličković, P., Cucurull, G., Casanova, A. et al. (2018). Graph attention networks. ArXiv171010903 Cs Stat 1710.10903v3: 1–12. 17 Landrum, G. (2016). RDKit: Open-­Source Cheminformatics Software. https:// github.com/rdkit/rdkit/releases/tag/Release_2016_09_4 18 Coley, C.W., Green, W.H., and Jensen, K.F. (2019). RDChiral: an RDKit wrapper for handling stereochemistry in retrosynthetic template extraction and application. J. Chem. Inf. Model. 59 (6): 2529–2537. 19 Hornik, K., Stinchcombe, M., and White, H. (1989). Multilayer feedforward networks are universal approximators. Neural Netw. 2 (5): 359–366. 20 Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep Learning. MIT Press. 21 Janet, J.P. and Kulik, H.J. (2020). Machine Learning in Chemistry. American Chemical Society. 22 Coley, C.W., Green, W.H., and Jensen, K.F. (2018). Machine learning in computer-­aided synthesis planning. Acc. Chem. Res. 51 (5): 1281–1289. 23 Schwaller, P., Vaucher, A.C., Laplaza, R. et al. Machine intelligence for chemical reaction space. WIREs Comput. Mol. Sci. n/a (n/a): e1604. 24 Strieth-­Kalthoff, F., Sandfort, F., Segler, M.H.S., and Glorius, F. (2020). Machine learning the ropes: principles, applications and directions in synthetic chemistry. Chem. Soc. Rev 49 (17): 6154–6168. 25 Corey, E., Long, A., and Rubenstein, S. (1985). Computer-­assisted analysis in organic synthesis. Science 228 (4698): 408–418.

  ­Reference

26 Szymkuc, S., Gajewska, E.P., Klucznik, T. et al. (2016). Computer-­assisted synthetic planning: the end of the beginning. Angew Chem Int Ed 55 (20): 5904–5937. 27 Genheden, S., Thakkar, A., Chadimová, V. et al. (2020). AiZynthFinder: a fast, robust and flexible open-­source software for retrosynthetic planning. J. Cheminformatics 12 (1): 70. 28 Thakkar, A., Kogej, T., Reymond, J.-­L. et al. (2020). Datasets and their influence on the development of computer assisted synthesis planning tools in the pharmaceutical domain. Chem. Sci. 11 (1): 154–168. 29 Roughley, S.D. and Jordan, A.M. (2011). The Medicinal Chemist’s Toolbox: an analysis of reactions used in the pursuit of drug candidates. J. Med. Chem. 54 (10): 3451–3479. 30 Schneider, N., Lowe, D.M., Sayle, R.A. et al. (2016). Big data from pharmaceutical patents: a computational analysis of medicinal chemists’ bread and butter. J. Med. Chem. 59 (9): 4385–4402. 31 Law, J., Zsoldos, Z., Simon, A. et al. (2009). Route designer: a retrosynthetic analysis tool utilizing automated retrosynthetic rule generation. J. Chem. Inf. Model. 49 (3): 593–602. 32 Heid, E., Liu, J., Aude, A., and Green, W.H. (2022). Influence of template size, canonicalization, and exclusivity for retrosynthesis and reaction prediction applications. J. Chem. Inf. Model. 62 (1): 16–26. 33 Segler, M.H.S. and Waller, M.P. (2017). Neural-­symbolic machine learning for retrosynthesis and reaction prediction. Chem. – Eur. J. 23 (25): 5966–5971. 34 Fortunato, M.E., Coley, C.W., Barnes, B.C., and Jensen, K.F. (2020). Data augmentation and pretraining for template-­based retrosynthetic prediction in computer-­aided synthesis planning. J. Chem. Inf. Model. 60 (7): 3398–3407. 35 Seidl, P., Renz, P., Dyubankova, N. et al. (2022). Improving few-­and zero-­shot reaction template prediction using modern Hopfield networks. J. Chem. Inf. Model 62 (9): 2111–2120. 36 Baylon, J.L., Cilfone, N.A., Gulcher, J.R., and Chittenden, T.W. (2019). Enhancing retrosynthetic reaction prediction with deep learning using multiscale reaction classification. J. Chem. Inf. Model. 59 (2): 673–688. 37 Ishida, S., Terayama, K., Kojima, R. et al. (2019). Prediction and interpretable visualization of retrosynthetic reactions using graph convolutional networks. J. Chem. Inf. Model. 59 (12): 5026–5033. 38 Coley, C.W., Rogers, L., Green, W.H., and Jensen, K.F. (2017). Computer-­assisted retrosynthesis based on molecular similarity. ACS Cent. Sci. 3 (12): 1237–1245. 39 Dai, H., Li, C., Coley, C. et al. (2019). Retrosynthesis prediction with conditional graph logic network. Adv. Neural Inf. Process. Syst. 32. 40 Chen, S. and Jung, Y. Deep retrosynthetic reaction prediction using local reactivity and global attention. JACS Au 1: 1612–1620.

455

456

11  Computer-­Assisted Synthesis Planning

41 Sun, R., Dai, H., Li, L. et al. (2021). Towards understanding retrosynthesis by energy-­based models. Adv. Neural Inform. Process. Syst. 34. 42 Lin, M.H., Tu, Z., and Coley, C.W. (2022). Improving the performance of models for one-­step retrosynthesis through re-­ranking. J. Cheminformatics 14 (1): 15. 43 Sacha, M., Błaż, M., Byrski, P. et al. (2021). Molecule edit graph attention network: modeling chemical reactions as sequences of graph edits. J. Chem. Inf. Model. 61 (7): 3273–3284. 44 Shi, C., Xu, M., Guo, H. et al. (2020). A graph to graphs framework for retrosynthesis prediction. Proc. 37th Int. Conf. Mach. Learn. 119: 8818–8827. 45 Somnath, V.R., Bunne, C., Coley, C.W. et al. (2021). Learning graph models for retrosynthesis prediction. arXiv 2006.07038v2: 1–15. 46 Yan, C., Ding, Q., Zhao, P. et al. (2020). RetroXpert: decompose retrosynthesis prediction like a chemist. Adv. Neural Inf. Process. Syst. 33: 11248–11258. 47 Wang, X., Li, Y., Qiu, J. et al. (2021). RetroPrime: a diverse, plausible and transformer-­based method for single-­step retrosynthesis predictions. Chem. Eng. J. 420: 129845. 48 Liu, B., Ramsundar, B., Kawthekar, P. et al. (2017). Retrosynthetic reaction prediction using neural sequence-­to-­sequence models. ACS Cent. Sci. 3 (10): 1103–1113. 49 Schwaller, P., Gaudin, T., Lányi, D. et al. (2018). “Found in Translation”: predicting outcomes of complex organic chemistry reactions using neural sequence-­to-­sequence models. Chem. Sci. 9 (28): 6091–6098. 50 Lin, K., Xu, Y., Pei, J., and Lai, L. (2020). Automatic retrosynthetic route planning using template-­free models. Chem. Sci. 11 (12): 3355–3364. 51 Lee, A.A., Yang, Q., Sresht, V. et al. (2019). Molecular transformer unifies reaction prediction and retrosynthesis across pharma chemical space. Chem. Commun. 55 (81): 12152–12155. 52 Duan, H., Wang, L., Zhang, C. et al. (2020). Retrosynthesis with attention-­based NMT model and chemical analysis of “wrong” predictions. RSC Adv. 10 (3): 1371–1378. 53 Irwin, R., Dimitriadis, S., He, J., and Bjerrum, E.J. (2022). Chemformer: a pre-­trained transformer for computational chemistry. Mach. Learn. Sci. Technol. 3 (1): 015022. 54 Zhu, J., Xia, Y., Qin, T. et al. (2021). Dual-­view molecule pre-­training. ArXiv210610234 Cs arXiv:2106.10234v2: 1–15. 55 Zheng, S., Rao, J., Zhang, Z. et al. (2020). Predicting retrosynthetic reactions using self-­corrected transformer neural networks. J. Chem. Inf. Model. 60 (1): 47–55. 56 Yoo, S., Kim, Y.-­S., Lee, K.H. et al. (2020). Graph-­aware transformer: is attention all graphs need? arXiv arXiv:2006.05213v1: 1–12. 57 Seo, S.-­W., Song, Y.Y., Yang, J.Y. et al. (2021). GTA: graph truncated attention for retrosynthesis. Proc. AAAI Conf. Artif. Intell. 35 (1): 531–539.

  ­Reference

58 Mao, K., Xiao, X., Xu, T. et al. (2021). Molecular graph enhanced transformer for retrosynthesis prediction. Neurocomputing 457: 193–202. 59 Tu, Z. and Coley, C.W. (2022). Permutation invariant graph-­to-­sequence model for template-­free retrosynthesis and reaction prediction. J. Chem. Inform. Modeling 62 (15): 3503–3513. 60 Chen, B., Shen, T., Jaakkola, T.S., and Barzilay, R. (2020). Learning to make generalizable and diverse predictions for retrosynthesis. arXiv arXiv:1910.09688v1: 1–11. 61 Kim, E., Lee, D., Kwon, Y. et al. (2021). Valid, plausible, and diverse retrosynthesis using tied two-­way transformers with latent variables. J. Chem. Inf. Model. 61 (1): 123–133. 62 Tetko, I.V., Karpov, P., and Van Deursen, R. (2020). State-­of-­the-­art augmented NLP transformer models for direct and single-­step retro synthesis. Nat. Commun. 11 (5575): 1–11. 63 Coulom, R. (2007). Efficient selectivity and backup operators in Monte-­Carlo tree search. Comput. Games 4630: 72–83. 64 Schwaller, P., Petraglia, R., Zullo, V. et al. (2020). Predicting retrosynthetic pathways using transformer-­based models and a hyper-­graph exploration strategy. Chem. Sci. 11 (12): 3316–3325. 65 Coley, C.W. (2019). ASKCOS Software Repository. https://github.com/connor coley/ASKCOS (accessed 1 June 2019). 66 Coley, C.W., Thomas, D.A., Lummiss, J.A.M. et al. (2019). A robotic platform for flow synthesis of organic compounds informed by AI planning. Science 365 (6453): eaax1566. 67 Chen, B., Li, C., Dai, H., and Song, L. (2020). Retro*: learning retrosynthetic planning with neural guided A* search. Proc. 37th Int. Conf. Mach. Learn. 1608–1616. 68 Heifets, A. (2014). Automated synthetic feasibility assessment: a data-­driven derivation of computational tools for medicinal chemistry. Thesis, University of Toronto. 69 Mo, Y., Guan, Y., Verma, P. et al. (2021). Evaluating and clustering retrosynthesis pathways with learned strategy. Chem. Sci. 12 (4): 1469–1478. 70 Genheden, S. and Bjerrum, E. (2022). PaRoutes: a framework for benchmarking retrosynthesis route predictions. Digital Discov. 1: 527–539. 71 Schneider, N., Stiefl, N., and Landrum, G.A. (2016). Whatś what: the (nearly) definitive guide to reaction role assignment. J. Chem. Inf. Model. 56 (12): 2336–2346. 72 Gao, H., Struble, T.J., Coley, C.W. et al. (2018). Using machine learning to predict suitable conditions for organic reactions. ACS Cent. Sci. 4 (11): 1465–1476. 73 Maser, M.R., Cui, A.Y., Ryou, S. et al. (2021). Multilabel classification models for the prediction of cross-­coupling reaction conditions. J. Chem. Inf. Model. 61 (1): 156–166.

457

458

11  Computer-­Assisted Synthesis Planning

74 Fitzner, M., Wuitschik, G., Koller, R. et al. (2020). What can reaction databases teach us about buchwald-­hartwig cross-­couplings? Chem. Sci. 11: 13085–13093. 75 Reizman, B.J. and Jensen, K.F. (2016). Feedback in flow for accelerated reaction development. Acc. Chem. Res. 49 (9): 1786–1796. 76 Shields, B.J., Stevens, J., Li, J. et al. (2021). Bayesian reaction optimization as a tool for chemical synthesis. Nature 590 (7844): 89–96. 77 Wei, J.N., Duvenaud, D., and Aspuru-­Guzik, A. (2016). Neural networks for the prediction of organic chemistry reactions. ACS Cent. Sci. 2 (10): 725–732. 78 Coley, C.W., Barzilay, R., Jaakkola, T.S. et al. (2017). Prediction of organic reaction outcomes using machine learning. ACS Cent. Sci. 3 (5): 434–443. 79 Jin, W., Coley, C., Barzilay, R., and Jaakkola, T. (2017). Predicting organic reaction outcomes with Weisfeiler-­Lehman network. Adv. Neural Inf. Process. Syst. 30: 1–10. 80 Coley, C.W., Jin, W., Rogers, L. et al. (2019). A graph-­convolutional neural network model for the prediction of chemical reactivity. Chem. Sci. 10 (2): 370–377. 81 Do, K., Tran, T., and Venkatesh, S. (2019). Graph transformation policy network for chemical reaction prediction. Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 750–760. 82 Qian, W.W., Russell, N.T., Simons, C.L.W. et al. (2020). Integrating deep neural networks and symbolic inference for organic reactivity prediction. ChemRxiv chemrxiv.11659563.v1: 1–24. 83 Bradshaw, J., Kusner, M.J., Paige, B., et al. (2019). A generative model for electron paths. International Conference on Learning Representations. 84 Bi, H., Wang, H., Shi, C. et al. (2021). Non-­autoregressive electron redistribution modeling for reaction prediction. Proc. 38th Int. Conf. Mach. Learn. 139: 904–913. 85 Nam, J. and Kim, J. (2016). Linking the neural machine translation and the prediction of organic chemistry reactions. ArXiv161209529 Cs arXiv: 1612.09529v1: 1–19. 86 Schwaller, P., Laino, T., Gaudin, T. et al. (2019). Molecular transformer: a model for uncertainty-­calibrated chemical reaction prediction. ACS Cent. Sci. 5 (9): 1572–1583. 87 ASKCOS. http://askcos.mit.edu/ (accessed 8 February 2019). 88 Coley, C.W., Rogers, L., Green, W.H., and Jensen, K.F. (2018). SCScore: Synthetic Complexity Learned from a Reaction Corpus. J. Chem. Inf. Model. 58 (2): 251–261. https://pubs.acs.org/doi/10.1021/acs.jcim.7b00622. 89 ASKCOS. tutorial. https://askcos.mit.edu/help/tutorial (accessed 31 March 2022). 90 Lowe, D. (2017). Chemical reactions from US patents (1976-­Sep2016). 91 Sterling, T. and Irwin, J.J. (2015). ZINC 15 – ligand discovery for everyone. J. Chem. Inf. Model. 55 (11): 2324–2337.

  ­Reference

92 Genheden, S., Thakkar, A., Chadimová, V. et al. (2020). AiZynthFinder: a fast, robust and flexible open-­source software for retrosynthetic planning. ChemRxiv 12: 1–9. 93 Conda —­Conda documentation. https://docs.conda.io/en/latest/ (accessed 31 March 2022). 94 Project Jupyter. https://jupyter.org (accessed 31 March 2022). 95 Chen, B. (2022). Retrosynthetic Planning with Retro*. https://github.com/ binghong-­ml/retro_star (accessed 31 March 2022). 96 eMolecules Free Downloads. https://downloads.emolecules.com/free/ (accessed 31 March 2022). 97 Klucznik, T., Mikulak-­Klucznik, B., McCormack, M.P. et al. (2018). Efficient syntheses of diverse, medicinally relevant targets planned by computer and executed in the laboratory. Chem 4 (3): 522–532. 98 Probst, D., Manica, M., Nana Teukam, Y.G. et al. (2022). Biocatalysed synthesis planning using data-­driven learning. Nat. Commun. 13 (1): 964. 99 Bøgevig, A., Federsel, H.-­J., Huerta, F. et al. (2015). Route design in the 21st century: The IC SYNTH Software Tool as an idea generator for synthesis prediction. Org. Process Res. Dev. 19 (2): 357–368. 100 Liu, C.-­H., Korablyov, M., Jastrzębski, S. et al. (2020). RetroGNN: approximating retrosynthesis by graph neural networks for de novo drug design. arXiv arXiv:2011.13042v1: 1–11. 101 Hadadi, N. and Hatzimanikatis, V. (2015). Design of computational retrobiosynthesis tools for the design of de novo synthetic pathways. Curr. Opin. Chem. Biol. 28: 99–104. 102 Koch, M., Duigou, T., and Faulon, J.-­L. (2020). Reinforcement learning for bioretrosynthesis. ACS Synth. Biol. 9 (1): 157–168. 103 Finley, S.D., Broadbelt, L.J., and Hatzimanikatis, V. (2009). Computational framework for predictive biodegradation. Biotechnol. Bioeng. 104 (6): 1086–1097. 104 Finnigan, W., Hepworth, L.J., Flitsch, S.L., and Turner, N.J. (2021). RetroBioCat as a computer-­aided synthesis planning tool for biocatalytic reactions and cascades. Nat. Catal. 4 (2): 98–104. 105 Levin, I., Liu, M., Voigt, C.A., and Coley, C.W. (2022). Merging enzymatic and synthetic chemistry with computational synthesis planning. Nat. Commun. 13: 7747. https://doi.org/10.1038/s41467-022-35422-y. 106 Lin, Y., Zhang, Z., Mahjour, B. et al. (2021). Reinforcing the supply chain of umifenovir and other antiviral drugs with retrosynthetic software. Nat. Commun. 12 (1): 7327.

459

461

Index a

ab initio methods  267 activation energies  261 affinity model  431 AiZynthFinder  429, 440–442 aldol reaction  262 AlphaFold  21 alternating polarity  88 AMBER. see Assisted Model Building and Energy Refinement (AMBER) amino acid dehydrogenase  20 aminolysis  20 Anaconda  352, 402 analysis (for flow)  134–135 analysis (for high‐throughput experimentation)  226, 358 analysis of variance (ANOVA)  180 analysis (statistical)  179 anode  82 array  199, 211–214 artifacts (in machine learning)  411 ASKCOS  435, 440, 449–450 aspartate aminotransferase  16

assay yield  227–228 Assisted Model Building and Energy Refinement (AMBER)  266 atom‐transfer radical addition  60 automation  229–232, 332, 378–381, 394, 449–450 autosampler  226–227 azoles  240

b

B3LYP  262, 271 Baeyer–Villiger monoxygenase  7 basis function  267 basis sets  262, 273, 276–278 limit  274 Bayesian optimization  375, 381, 399 biocatalysis  6–15, 352 Boltzmann weighted average  316, 328 Born–Oppenheimer approximation  267 Box–Behnken design  169 Buchwald‐Hartwig amination  355, 362, 406 buffers  11

Enabling Tools and Techniques for Organic Synthesis: A Practical Guide to Experimentation, Automation, and Computation, First Edition. Edited by Stephen G. Newman. © 2023 John Wiley & Sons, Inc. Published 2023 by John Wiley & Sons, Inc.

462

Index

c

CALB. see Candida antarctica: lipase B (CALB) Candida antarctica, lipase B (CALB)  14, 20, 28 Cartesian coordinates  286 cascade  2 cathode  82 center composite inscribed (CCI) design  169 CHARMM. see Chemistry at HARvard Molecular Mechanics (CHARMM) C–H arylation  376, 406 Chematica  425 Chemdraw  439 cheminformatics  426 Chemistry at HARvard Molecular Mechanics (CHARMM)  266 chemoselectivity  143 Chemputer  379 Chemspeed  382 ChemStation  229 CHESHIRE Chemical Shift Repository  317, 330 C–H functionalization  240 chromatography  128, 226 chromophore  39 C–N coupling. see Buchwald–Hartwig amination compact fluorescent lamps (CFLs)  51, 64 competitive sequential reactions  112–114 computational chemistry  259–306, 261, 314 computational cost  266 computational methods  265–273

computational power  261 computer‐aided synthesis planning (CASP)  424–425 Computer‐Assisted Structure Elucidation (CASE)  332 configuration interaction  268 conformational analysis  285–286, 318 Conformer–Rotamer Ensemble Sampling Tool (CREST)  285 confounder  165 Corey–Bakshi–Shibata (CBS) reduction  3 coupled‐cluster theory  268 coupling constants (calculating)  330 Crabtree’s catalyst  8 cross‐validation  404 crystallization  357 current  85, 100 curse of dimensionality  403 cycloaddition  42, 63 cyclobutanes  57

d

databases  364, 369, 436 data privacy  448 visualization  228–229, 351 wrangling  374 dearomatization  59 decatungstate anion. See TBADT deep highway networks  431 deep learning  336, 428 definitive screening designs (DSDs)  166 dehydrogenase  15 density functional theory (DFT)  265, 269–271, 375 calculations  355

Index

deoxyfluorination  360 descriptors  367, 400 Design Expert  193 Design of Experiments (DoE)  151, 397 DFT. see density functional theory (DFT) dielectric constant  358 Diels–Alder cycloaddition  287 diffusion  111 digitization  379 directed evolution  7 disguised chemical selectivity  113 dispensing liquids and solids  221–223 dispersion  271, 324 dynamic kinetic resolutions  4

e

effective core potentials  278 ElectraSyn 2.0  80 electrochemical cell  81 electrochemistry  117, 215 electrode  82–83, 94, 100 electrolysis  77 electrolyte  93, 95 electron transfer  45 electrosynthesis  77, 78 enzymes  1, 446 engineering  24–25 unit  15 enzyme commission (EC) number  15 error (experimental)  156–157, 175 Escherichia coli  13 excited state  39 exothermic reactions  112 explicit solvation  321

exploitation mode (for optimization)  404 exploration mode (for optimization)  404 extremophiles  10 E/Z isomerization  55

f

face–centered composite design  168 factor  152, 171–173 settings  157 factorial design  159 falling film reactor  53 features. see descriptors fittings  121 flow biocatalysis  21–24 flow chemistry  52–53, 107–144, 109, 248, 449 flow photochemistry reactors  53 flow rate  110 flow reactor  120, 140 fluorinated ethylene propylene (FEP) and fluorinated ethylene‐ propylene copolymer (FEP)  53, 62 fractional factorial design  165

g

Gaussian  264, 316, 356 Gaussian input files  288 Gaussian‐type orbitals (GTOs)  274 generalized gradient approximation  270 genetic engineering  9 geometry optimizations  281–282, 291–294, 315, 323 GitHub  350, 385, 409 GL Science  121

463

464

Index

glycosidase  21 graph convolutional networks  431 graph‐edit‐based retrosynthesis methods  432 greedy. see exploitation mode (for optimization) green chemistry  39, 368 Grotthuss–Draper law  39

h

Hamiltonian operator  267 Hartree–Fock  265 Harvard Apparatus  121 heatmap  229 heavy atom effects  329 heterogeneous catalysis  143 heterogeneous reaction mixtures  200 high‐performance liquid chromatography (HPLC)  128, 178, 382 high‐throughput experimentation  199, 357, 365 HOMO  39 Horner–Wadsworth–Emmons olefination  406 horseradish peroxidase  19 HPLC. see high‐performance liquid chromatography (HPLC) HPLC pump  62 hydroalkylation  65 hydrogen atom transfer (HAT)  45, 63 hydrolase  16–17

i

ICSynth  445 IKA  80 imides  90 imine reductases  4 immobilization (of enzymes)  22

immobilized enzymes  13 implicit solvation  324 incandescent lamp  50 Inchi  353 indoles  59 inertion box  221 Integer Linear Programming  438 integration grid  302 interaction (statistical)  153, 160 internal scaling factors  329 internal standard  94, 128, 227–228 intersystem crossing  40 intrinsic reaction coordinate  285 ionic strength  10–11 isomerase  18 isotropic shielding constants  324, 329

j

JMP  193 Jupyter notebook  354

k

katal  15 Kessil  51 ketones  41, 90, 114 kinetic resolutions  20, 27, 28 KitAlysis  221, 232

l

LabMate.ML  395 lab‐on‐a‐chip  53 Lambert–Beer law  41 laminar flow  111 level (statistics)  152 level of theory  323–324. see also model chemistry ligands  191, 235, 243–244, 364, 384 ligase  18

Index

light‐emitting diodes (LEDs)  51 light sources  48–52 linear regression model  412 local density approximation  270 low‐pressure mercury arcs  49–50 LUMO  39 lyase  17 lyophilized enzymes  13 lysozyme  19

m

machine learning  249, 336, 355, 394, 426 main effect  159 Markov decision processes  438 mean absolute error  405 medium‐pressure mercury arcs  50 micromixer  111, 119 microreactors  53, 109–110, 120 miniaturization  203 Minitab  193 Minnesota functionals  272 Mitsunobu reaction  377 mixing  111–112 Modde  181 model chemistry  262, 316 model (mathematical)  155 molecular dynamics  266, 319 molecular graph  426 molecular mechanics  266 molecular representation  426–428 Møller–Plesset perturbation theory  268 Monte Carlo sampling  319 Monte Carlo simulation  369 Monte Carlo Tree Search (MCTS)  434, 448 Morgan fingerprint  426

multiphase reactions  117–118 Multiple linear regression (MLR)  163 multiwell plate  199, 214 muramidase. see lysozyme

n

natural products  336 N‐chloroamines  60 neural networks  334, 426, 437 Ni catalysis  92, 240 NMR. see nuclear magnetic resonance (NMR) Norrish type I reaction  42, 43 Norrish–Yang reaction  43 nuclear magnetic resonance (NMR)  263, 315 prediction  313 nucleophilic substitution  158

o

one hot encoding  402 one variable at a time (OVAT)  151 Open Reaction Database  364 OPLS. see Optimized Potential for Liquid Simulations (OPLS) optimal design  176 optimization  203–204 Optimized Potential for Liquid Simulations (OPLS)  266 organocatalyst  55 organolithiums  125, 136 oxoreductase  15 ozone  54

p

pandas  352, 401 parameter. see factor Pareto set  399

465

466

Index

Paternò–Büchi reaction  43, 58 Pd catalysis  355, 364 PEEK  122 perfluoroalkoxyalkanes (PFA)  53 photobiocatalysis  25 photocatalyst  45 photochemistry  39, 117 photocycloaddition  57 [2+2] photocycloaddition  43 photoheterolysis  44 photohomolysis  44 photoreaction  48 photoredox catalysis  51, 63, 215 photosensitizer  44 pipettes  224 Pistachio database  447 plate. see multiwell plate polar bear chiller  218 porcine pancreas lipase (PPL)  28 potential (electrochemical)  85 potentially significant factors (PSFs)  171 principal component analysis (PCA)  169, 191 process mass intensity  368 programming  347, 349 languages  350, 359, 371, 401 PTFE  121 PubChem  364 pumps  120–121 Python  352, 371, 401

q

quantum mechanics  262, 319 quartz lamp  57 tubes  49

r

random forest algorithm  362, 402 RDKit  427, 431 reaction blocks  214 discovery  240–243 integration  114, 139–141 optimization  167, 210–211, 374–378, 381–384, 397 rule (see template (reaction)) reactors (for high throughput experimentation)  214–220 Reaxys  25, 436 Reaxys database  429 reductive amination  407 reductive cross‐coupling  90 reference electrode  82 reproducibility  157 residence time  110, 123 residual chemical shift anisotropy  333 residuals normal probability plot  181 resistance  86 response  152 response surface  404 response surface design  167–169 Retro  442–443 RetroBioCat  21 retrospective evaluation  409 retrosynthesis  428–436 robotics  382. see also automation R programming language  358 RXN  445

s

sacrificial electrode  100 scale‐up (in flow)  118

Index

scaling factors  317, 329 Schrödinger equation  267 SciFinder  25, 445 SciKitLearn  359, 401 screening design  164–167 self‐consistent field (SCF)  267 self‐optimizing reactions  191 semiempirical computational methods  269 single electron transfer (SET)  63 single‐point calculation  281 singlet oxygen  45, 62 singlet state  40 Slater‐type orbitals  274 SMARTS  427 SMILES  353, 426 SN2 reaction  283 software (for data analysis)  179 solid dispensers  222 solubility  358 solubilized enzymes  13 solvent models  301, 321 solvents (for biocatalysis)  10 solvents (for electrochemistry)  93, 97 solvents (for photochemistry)  45–47 Spartan  355 specific activity  14 spin state  40 Spotfire  229, 351 stabilizing time. see steady state Stable Noisy Optimization by Branch and Fit (SNOBFIT)  397 standard activity assay  27 standard deviation  362 standard state  302–303

statistical analysis  351 steady state  124 stereoelectronic features  237 sunflow reactor  61 supercomputer  264 Suzuki–Miyaura reaction  233, 365, 382, 413, 427 Synthia  425, 444, 450–451 syringe  120–121, 126 systematic dihedral angle sampling  319

t

TBADT  65 Teflon. see PTFE telescoping. see reaction integration template (reaction)  427, 429 text extraction  354–357 thermal correction  284 toxicity of enzymes  13 training (machine learning models)  361 transesterification  14 transferase  16 transition state (identification of)  282, 294–297 translocase  18 triplet state  40 trypsin  20 tungsten‐halogen lamp  50 turbulent flow  111

u

Ugi reaction  406 UNIX  401 unpaired electrons  303–304 USPTO dataset  431

467

468

Index

UV‐A  51 UV‐B  54 UV‐C  54 UV‐visible spectrum  45

v

variable. see factor

vitamin D  56 voltage  79

w

wave function  267 whole cells  12 Wittig reaction  170